<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Sandeep VM on Medium]]></title>
        <description><![CDATA[Stories by Sandeep VM on Medium]]></description>
        <link>https://medium.com/@svm161265?source=rss-9298123e8771------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*C3mhWWwJFJVQ9p5Zbx5XmQ.jpeg</url>
            <title>Stories by Sandeep VM on Medium</title>
            <link>https://medium.com/@svm161265?source=rss-9298123e8771------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 07:02:42 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@svm161265/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Invisible Guardian: How Error Control Decoders Keep Our Digital World From Falling Apart]]></title>
            <link>https://medium.com/@svm161265/the-invisible-guardian-how-error-control-decoders-keep-our-digital-world-from-falling-apart-5f0ec167ea11?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/5f0ec167ea11</guid>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Sun, 12 Jul 2026 20:38:29 GMT</pubDate>
            <atom:updated>2026-07-12T20:38:29.535Z</atom:updated>
            <content:encoded><![CDATA[<p>The Invisible Guardian: How Error Control Decoders Keep Our Digital World From Falling Apart</p><p>The Silent Miracle of Modern Communication</p><p>Every second, billions of bits of data hurtle across the globe, pulsing through transcontinental fiber-optic cables and bouncing off satellites to reach our screens. Whether it’s a 4K video stream or a secure banking transaction, we have come to expect absolute digital perfection. Yet, the physical medium through which this data travels is inherently hostile. Between the sender and the receiver lies a gauntlet of electromagnetic interference and physical degradation. How does this vast ocean of information arrive without a single mistake?</p><p>The answer lies with an unsung hero of the digital age: the Error Control Decoder. This isn&#39;t just a piece of software; it is a mathematical gatekeeper that ensures the integrity of our digital lives, working silently in the background to make sense of the chaos.</p><p>The Dual Mission: Detection Meets Correction</p><p>To understand the Error Control Decoder, we must look at its two-pronged strategy for maintaining data integrity. In the world of engineering, there is a massive conceptual leap between knowing something is wrong and having the power to fix it.</p><p>First, the system must detect errors. Think of this as a digital smoke alarm. During the diagnostic phase, the decoder scans the incoming bitstream to identify if any information has been corrupted during transit. In a basic &quot;detection-only&quot; scenario, the system realizes the data is broken and must discard it, forcing the sender to re-transmit the entire packet. This is functional, but slow and inefficient.</p><p>The second, more sophisticated mission is to correct errors. This is where the magic happens. A high-performance decoder doesn&#39;t just flag the &quot;broken&quot; data; it uses internal logic to reconstruct the original information on the fly. By correcting errors at the destination, the decoder eliminates the need for time-consuming re-transmissions, transforming a fragile, stuttering connection into a robust, seamless stream of data.</p><p>The Power of One: The Hamming Code Breakthrough</p><p>To appreciate how these mathematical guardians function, we can look at one of the most elegant solutions in communication science: the Hamming code. This protocol is a masterclass in efficiency, utilizing the concept of &quot;redundancy&quot;—adding a few extra bits to a message that act as a roadmap for the decoder. These extra bits don&#39;t carry the message itself, but they provide the mathematical landmarks necessary to navigate back to the truth if a bit gets lost.</p><p>The Hamming code represents a specific, powerful milestone in our ability to maintain data integrity:</p><p>&quot;Hamming code can correct single bit error.&quot;</p><p>Correcting a single bit might sound like a minor achievement, but in a world of binary logic, it is everything. A single flipped bit—a &quot;0&quot; becoming a &quot;1&quot;—can be the difference between a clear pixel and a glitch, or a valid command and a system crash. By identifying exactly which bit was flipped and reversing it, the Hamming code ensures that minor interference doesn&#39;t escalate into a total communication failure.</p><p>The Ultimate Goal: Building a More Reliable Connection</p><p>The complex interplay of detection and correction leads to one definitive result: more reliable communication.</p><p>The Error Control Decoder acts as a vital safety net against the &quot;noise&quot; of our physical reality. In a server farm, this might be thermal fluctuations; for a satellite, it could be a solar flare; and for your home Wi-Fi, it might be the microwave in the kitchen. Without these decoders, our digital world would be a mosaic of pixelated errors and dropped calls. By cleaning up the interference inherent in the physical world, these decoders create the stable, high-fidelity environment we now take for granted.</p><p>The Future of Flawless Data</p><p>As we push into the next frontier of Digital Communication and Compressive Sensing, the importance of these decoders is only magnifying. Compressive sensing, in particular, relies on the ability to reconstruct high-resolution data from very few samples. This efficiency makes the system incredibly fragile—if those few samples are corrupted, the entire reconstruction fails. Here, the Error Control Decoder is the essential foundation that makes high-efficiency sensing possible.</p><p>As our hunger for data grows and our networks become more complex, we become ever more dependent on these invisible safety nets. It raises a fascinating question: in an age where our global infrastructure rests on the flawless transfer of bits, how often do we stop to consider the invisible math keeping the world connected?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5f0ec167ea11" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SAR Preparation for NBA- Tier II Engineering Institutes]]></title>
            <link>https://medium.com/@svm161265/sar-preparation-for-nba-tier-ii-engineering-institutes-c3bc3538d06d?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/c3bc3538d06d</guid>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Mon, 02 Oct 2023 05:00:56 GMT</pubDate>
            <atom:updated>2023-10-06T09:57:38.405Z</atom:updated>
            <content:encoded><![CDATA[<h4>According to the Washington Accord, signed by India on 13th June 2014, the engineers graduating from Indian universities are universally recognized if and only if they pass out from the institute that is accredited by the National Board of Accreditation (NBA). All the students passed out from an NBA accredited institutions need not go through rigorous filtering process in all the countries that have signed the accord, either for higher studies or for employment. In this regard, it has become almost mandatory for the institutions to be accredited by NBA to prove the quality of their education to be at par with the International standards.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/728/1*eoYCcPQad1_tmiR0GQhxPw.jpeg" /></figure><p>NBA has different guidelines for Tier I and Tier II Engineering Institutes. Here let’s concentrate on Tier II institutions. The accreditation is granted at program- level, not at Institute-level.</p><p>The NBA process reinforces Outcome Based Education(OBE) system. In this system, the education quality of the program is evaluated on continous basis to ensure that the quality goes monotonously upward with time.</p><p>The NBA accreditation is a 3 years process. Once a program plans for accreditation, it must start working on OBE guidelines. All the activities of the program over the entire year is continously monitored. SWOT analysis is conducted. This analysis provides us our strengths, weaknesses that will help us in identifying the opportunities and threats we have in the coming future. The appropriate actions for encashing the opportunities and to overcome the threats are identified. These actions are then properly enforced in the next year. This guarantees monotonous improvement of the program’s quality. The quality is measured in terms of opportunities provided to the students for their employment, higher studies &amp; research and in becoming entrepreneurs. At the end of 3 years, the Program can apply for NBA accreditation.</p><p>The Self Assessment Report (SAR) is an essence of the activities carried out by the Program in the past 3 years to improve the competence of the students passing out from the Program for their employability, admissions to higher studies or to become entrepreneurs in any of the countries that have signed the accord.</p><p>The SAR contains 10 Criteria of which 7 discuss about the Program and the remaining 3 deal with the institute.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*5zzPh7jQJRllT7KctXJ6rQ.jpeg" /></figure><p>Let’s now go through these 7 Criteria in detail:</p><h4><strong>Criterion 1: Vision, Mission and Program Educational Objectives(PEOs):</strong></h4><ul><li>Vision is what we want to become in the next few years down the line. The vision is the dreams we see for our alumni, maybe 10 years, after they leave our institution.</li><li>Mission is all about what we’re today. It’s about where we want to see our students few years after completing their Program.</li><li>Strategy is about how we achieve the vision. These are the mechanisms we need to apply to achieve these visions.</li><li>Goals and objectives help us in measuring how successful we are in achieving the vision.</li></ul><p>The Program must have it’s own Vision, Mission, PEOs and Program Specific Outcomes (PSOs). The Vision, Mission of the Program must be consistent with that of the institute.</p><p>The Vision, Mission and PEOs must be decided taking all the stakeholders, like the faculty members, the parents, industry and alumni, into confidence.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*-82CN3Zp2lby4781JeIeBA.png" /></figure><p>Once the Vision, Mission and PEOs are defined, they must be published and disseminated to all the stakeholders. This will help us to be serious in achieving our vision.</p><h4>Criterion 2: Program Curriculum &amp; Teaching Learning Process:</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/1*fdKBpJyEm3sMkOlfd3vkmA.png" /></figure><p>The next issue is to understand the relevance of the curriculum with the current needs of the stakeholders. For Tier II institutions, the curriculum is decided by the university to which the institution is affiliated. The changes in the needs of the stakeholders and the curriculum never sync. This creates gaps between the curriculum and the expectations of the stakeholders from our students. The gaps identified from the past experience must be filled. The Program Outcomes (POs) defined by the NBA help us in identifying the gaps in the curriculum. The curriculum must ensure to fulfill all the POs. In case any PO is weakly fulfilled, it creates a gap.</p><p>If major changes are needed in curriculum, we can request the Board of Studies (BoS) of the university to upgrade the curriculum.</p><p>For minor changes, the Program can arrange special events to fill these gaps. The events can be Workshops, Seminars, Case-studies, Internships, Invited lectures, etc.</p><p>The next issue is about the Teaching-Learning process. This deals with the various methods adopted by the faculty members to teach the students. The first activity in this process begins with defining the academic calendar for the semester. The calendar highlights the data on which various events, like classwork, committee meetings, parent meetings, internal examination dates etc., are scheduled. Everyone in the Program must adhere to this schedule to ensure high quality outcome.</p><p>In order to make sure that each and every student in the class understands what’s being taught, various innovative methods must be adopted by the faculty members. In this criterion, it’s expected to highlight all the innovative techniques used to enlighten the students with the knowledge required for sustained employability.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/732/1*7ldrtIOc3h_ny8q1QxzFtg.jpeg" /></figure><p>The knowledge must be reinforced through practical workouts by the students. Practicals are given higher priority over the theory in OBE system.</p><p>During the vacation periods, the students are encouraged to take up internships and understand the real-time scenario at the industrial sector.</p><p>Finally, all the knowledge gathered over the 3 years of engineering must be applied to build or create through the project work in their final year of the curriculum. In this phase, a group of students work together to create something useful to the society. During this phase, they</p><ul><li>Learn the importance of each course they studied</li><li>Develop innovative skills</li><li>Learn to work in groups</li><li>Learn to solve any given problems by themselves</li></ul><p>Continuous evaluation is the core of OBE system. The students are subjected to continue evaluation through various methods like Continuous Internal Evaluation(CIE), conducting quiz in regular classes, assignments, making students to deliver seminars in selected topics etc. The quiz will help the faculty to know the depth of understanding of each student in the class. The future lectures may be reoriented to improve the students&#39; performance. To reinforce the learning, additional problems may be asked to solve at home and submit as assignments. One can make the students to deliver seminars on the topic beyond syllabus that can be useful in reinforcing the understanding. All these activities must ensure good quality and help in further improving knowledge and must have relevance to the COs of the course.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/489/1*1KooDv9iVJiISFvOBYE6iw.gif" /></figure><p>The CIE analysis helps the faculty members to understand the academic status of each student in the class. This analysis helps in identifying academically weak students for whom we can give additional attention and guide the weak students to improve their performance. The analysis also allows to identify bright students and we can encourage them to participate in various competitions and show their talents and get recognised. These recognitions help these bright students to build strong CV.</p><p>All the above are possible only if the evaluation process is effective and of good quality. The Teaching-Learning process must ensure the quality of the question papers set for the internal exams. These papers must follow the Revised Bloom’s Taxonomy and the pattern must be similar to the question papers of the university examinations. The internal examination question papers must ensure the coverage of all the course outcomes.</p><p>The whole process is reinforced through the feedback system. Here the feedback is regularly collected from the students, their parents, alumni, their employers etc. This system will help one to know their strengths and weaknesses. The feedback thus collected help to monotonously improve the academic quality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/1*IuZtOWExSZjWOBwov1sqfg.png" /></figure><p>Teaching-Learning process plays a major role in deciding the quality of the project works of the students. The process begins with creating project batches. The team members of the batches are asked to discuss among themselves to decide the domain in which to proceed further. Once the domains are decided, the guides are allotted considering the guide’s expertise. The team members in coordination with their guide review the latest literature and the target title for the project is decided and submitted to the projects coordinator. The domain and the title are selected such that the projects help in attaining the POs and PSOs. A committee of senior faculty will verify and suggest the changes, if any, and then approve the titles. After the title approval, the team with their guide dig deep into the literature to have more insight about the topic selected. The problem at hand is fully understood and then the solution is designed. This solution is them implemented and tested for its correctness. The process is repeated multiple times till everything works perfect. The whole process of project building is continously evaluated to ensure the progress. Then the whole story of building the project work is well documented as the thesis. The project is exhibited to the committee and the authorized thesis is submitted. The project work gives an opportunity to the students to showcase their talents as individuals as well as a team. Finally the quality of the project can be evaluated upon the publications that come out as a result of the project work. The guide must encourage the team members to actively participate at each stage of the project work and ensure that the quality of the work guarantees quality publications.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UiaIS9imaNY0Gcx8j6zk3A.png" /></figure><p>The knowledge gathered by the students are further reinforced through internships at industrial environment. It helps the students to understand and inculcate the work culture at the industries well in advance.</p><h4>Criterion 3: Course Outcomes(CO) and Program Outcomes(PO)</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gNnqkRGN0xetMQ8kHN67cw.png" /></figure><p>The course outcomes must be planned well in advance for each course in the program. The COs will define how the course will help in attaining the goals of the program.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IKaP3_srFGrD4aeQO017Mw.jpeg" /></figure><p>Each CO of the course is mapped to the POs and PSOs with appropriate rank. Blank for no relevance, 1, 2,and 3 for low, medium and high relevance respectively. This will help one to know the relevance of each course in the Program and what all POs it fulfils. The curriculum is supposed to be covering all the POs and PSOs. Looking at the CO-PO/PSO mappings of all courses of the program, we may find some gaps in fulfilling the POs and PSOs. If these gaps are small, they can be addressed at Program level by arranging activities beyond curriculum. If the gaps are huge, then a requisition may be forwarded to the Board of Studies of the affiliated university suggesting to include relevant topics into the future curriculum at appropriate places.</p><p>At the end of each semester, the performance of the students must be evaluated. Here, the attainment level of each PO is compared with the target level decided at the beginning of the course. The criterion 7 will use this information to identify the gaps in Teaching-Learning process, and suggest activities to fill these gaps in the forthcoming year. The sole propose is to ensure monotonous improvement in the performance test after year.</p><h4>Criterion 4: Students ‘ performance</h4><p>This criterion compares the performance of the is the students year after year.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i7gtKjDtUm4AD4Im7Z1Epw.png" /></figure><p>It all begins with indexing of Institute in general and our Program in particular at local and global levels. The enrolment rate and the quality of the students who wished to join our program is a clear indicator of the quality of our program.</p><p>The next parameter in proving our quality is through the success rates of our students. This rate can be measured in two ways. The student who never fails anywhere in the way from the first year till the completion of the program is a great at to the Program, Institute, as well as to the society. The percentage of such students will surely one of the best indicators of the quality of the Program. There is another set of students who, due to some reasons, might have failed somewhere on the way in completing their Program, but are brave enough to complete the Program in time, i.e. in four years from their admission date. For such students, ill-health, socio- economic problems etc., are some usual problems hindering their path towards success. These students do need recognition and hence are another indicator of our Program’s quality.</p><p>All the levels of the Program must be given equal emphasis. The performance analysis of the students at each year of the Program must be carefully studied to ensure the consistent improvement in performance by our students at every level.</p><p>The important assessors of the quality of the Program are our Stakeholders — employers, other universities and mainly the society. Quantitatively these can be measured in terms of placements, admissions to higher studies and becoming entrepreneurs respectively.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/509/1*kuYOOdz5jM_00ynI7FBbvA.jpeg" /></figure><p>The number of companies visit the campus for placements is the best indicator of how much the companies appreciate the quality of the Program in general and the students in particular. The higher the number of students placed and the higher the packages they are offered, the higher the Program is recognized. The stakeholders, at the institute level as well as at Program level, must ensure monotonous improve of this parameter through proper training activities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/765/1*06RhCHvX0RCHZzNtdO9trw.jpeg" /></figure><p>When the students want to continue their studies, they need to compete with the students from all the institutions. There will be tough competition and they need to score high in exams like GATE, GRE, TOEFL etc. and hence can be a good marker of quality of the Program and institution at national and international levels. One must always encourage the students to go for higher studies at premier institutions in the country or abroad.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*4sNyn8dVP8FWKmAgVo6Jxg.jpeg" /></figure><p>Every batch of students will surely consist of some who dream to be self- employed. They want to become entrepreneurs by creating their own world. If these students are properly trained during the program, that can change the world through their dreams. The acceptance of their dream by the society depends upon the quality of the Program and hence the entrepreneurship imbibed into the minds of such students. The more the number of students recognized as entrepreneurs, the higher Program and the institution be recognized by the society.</p><p>The students can be made globally competent by encouraging them to be part of professional societies such as, IEEE, IETE, ACM etc. Regular activities at the program and institution levels through these societies update the students to the current needs of the industry and society. The quality of the Program and the institute keep improving through these activities and this in turn enhances the opportunities for our students.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*u0gHsc7rB1guXnm1kHCCRg.jpeg" /></figure><p>The creativity in the students can be brought to light by allowing them to publish their ideas. The platforms like magazines and news letters published at Program and institution level can provide these opportunities. These publications can further motivate others and lead a path for others to excel too. This helps in continously enhancing the skills and performance of every student in the Program and the institution.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/440/1*-OM4ZvJkGLEIwCnSsFk7WA.png" /></figure><p>The overall performance of the students can further be improved by exposing them to the world will in advance so they can make SWOT analysis and act upon in the direction that help them in improving their strengths and working upon their weaknesses. Encouraging the students to participate in inter- institute events that are organized at local, state, national and international levels will help them to compare themselves with others and cultivate best practices identified during such events. Through these events the quality of the Program as well as the institute can be publicized to the stakeholders which further improves the opportunities to our students.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/1*lxifl4tRzBMLQe9FAPNQvw.jpeg" /></figure><h4>Criterion 5: Faculty Information and Contributions</h4><p>The faculty plays very important role in the improving the quality of the institute. The number of qualified and experienced faculty is an important asset of the intuition as well as the Program. The more the faculty, the more efficiently the workload is distributed and hence better quality of teaching. The higher the qualification of the faculty, the better will be their teaching quality. The wider the spectrum of specializations, the better every aspect of the Program can be envisioned. These factors ensure better academic ambience and help in producing efficient and excellent engineers.</p><p>The Student-Faculty Ratio(SFR) gives the quantitative information about how many students are to be handled by any faculty in the Program. The lower is this ratio, the better the faculty can concentrate in providing quality education to the students. Hence SFR is another important parameter in judging the education quality of the Program and hence the institute.</p><p>The knowledge to the students must be imparted at various levels depending upon the complexity of the courses. This must cover all the levels of Revised Bloom’s Taxonomy (RBT) from remembering to design. It makes the curriculum to be designed appropriately to cater all these factors. From the faculty side, we would require the faculty with various capabilities- some to lecture, some to help students conduct experiments to reinforce what is learnt in classes, some to tinker curious minds to explore deeper into the courses and some to help the students to take up the research. The entry level faculty (Assistant Professor) help in regular class work and laboratory sessions and the mid level faculty (Associate Professor) deal with the curious mind of the students whereas the top level faculty (Professor) indulge in research activities to encourage students as well as the faculty. For success of any program, all these faculty are essential, though the ratio is different. This cadre ratio of any program is parameter in deciding the Program ’s academic quality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f9t0-ya22rlTtuWP2A2nIA.jpeg" /></figure><p>It is very much evident from various studies that the higher the faculty is qualified, the better will be the teaching quality. This is due to the very fact that the wider base of the knowledge, the more examples are given to emboss the understands into the minds of the students. The number of faculty with higher qualifications are an asset to the Program as well as the institute. This parameter is another important pointer to the Program’s quality.</p><p>Another important factor that plays an active role in deciding the Program’s quality is by the satisfaction level of the faculty. This directly influences upon the quality of participation of the faculty in all the phases of the Program to impart the knowledge to the students. The duration any faculty would continue to be associated with the Program depends upon this satisfaction level. The higher the satisfied the faculty is, the longer the association can be expected. This the retention rate of the faculty is yet another metric to the quality of the program.</p><p>The usual chalk and board approach of teaching is no longer sufficient to engrave the knowledge upon the minds of the students. The faculty needs to look for innovative ideas to improve the teaching quality. The Program with the faculty indulged in more innovative techniques to impart knowledge will surely succeed in achieving its goals.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*u1bwd_cCVnpT4luZH9wcRQ.png" /></figure><p>The faculty must update their knowledge regularly to match the needs of the industry. This compulses the faculty to particulate in regular training sessions. Regular training programs are arranged by various institutions to meet this requirement. The Program and the intuitions must encourage all the faculty to attend these training programs on regular basis. The more faculty of the Program are involved in these activities, the wider will be the knowledge and hence the better will be the academic quality. Hence it is an important metric of Program’s quality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/753/1*dBA422YYOcc-OxF_o0J_4w.jpeg" /></figure><p>Research is an important component of a faculty’s life. It ensures monotonous improvement in the knowledge bank and hence the quality of the Program. The research involves in unearthing new concepts and techniques to implement these new concepts. Publications are the integral part of research. The research outcomes can be published to the world through conferences and journals. The quality of the journal is decided by its impact factor. The higher is the impact factor, the larger the audience the publication can reach. The possibility of any research being published in journals with high impact factors depends upon various factor like originality of the work, quality of research, it’s impact on the society and it’s effective applications. The number of such publications by the faculty represents the strength of the Program. The regularity of such publications by the faculty is another measuring yard of the Program’s quality.</p><p>The knowledge grows when we impart out knowledge to others. The quantitative measure of this growth can be the number is research scholars one is mentoring. The more the faculty of the Program mentor others the faster the knowledge bank of the Program grows. This number can be a better measure of performance of the Program.</p><p>The innovative ideas with the faculty can become reality after the experimentation and building the prototypes. This process may need resources with higher budget. This budget always may not be possible to bear by the Program or the institute. In such cases, research institutions/ organizations/ agencies come to rescue. They can sponsor the research. These sponsorships are available only to the quality projects under the able guidance well qualified to handle such projects. The number of sponsored projects in the program reflects the ability of the faculty and hence the quality of the Program.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7vpHyNlsTGg8u0no6nkzRA.png" /></figure><p>The consultancy services provided to the industry by the faculty is another feather in the cap of the Program. This industry- institute interaction can be a multifold benefit to the faculty as well as the program.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/602/1*CF5pvwYlOmus6DhQX99xTw.jpeg" /></figure><p>The recognition of the faculty in the Program and institute level can be a driving force for any faculty to give to the Program beyond the capacity and hence efficient results cab be expected. Regular faculty appraisal will boost the morale of the faculty and hence allow monotonous improvement in their performance. A well defined appraisal system will prove the attitude of the institute towards maintaining the quality of the academic system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/824/1*gEUZInQrInw5hv79-4yTeQ.png" /></figure><h4>6 Facilities and Technical support</h4><p>The facilities provided by the Program to the faculty and the students will play a major role in building a quality education system. The basic facility for an engineering program is the laboratories. The Program must have sufficient number of labs to conduct all the lab courses in the program. All the labs must be enriched with necessary equipment/ machines/ components/ software for smooth conduction of all the experiments ensuring that each students will get hands on experience. The students must get all the opportunities to learn the courses to the root. The facility availability is another metric to intend of the Program to well train their students.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/626/1*ZEu_sa_tnIIjv1i4gIS56g.jpeg" /></figure><p>The next important concern is about the manpower provided to take care of the laboratories. It can be the technical staff or the maintenance staff. Sufficient number of the technical staff will ensure proper skill transfer to the students while experimenting on any concept. Maintenance of the equipment/ systems in the lab is equally important for smooth running of the lab course. The number of these staff is a first indication of the intent of the Program towards quality education.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/900/1*63z-mAtr5V5826P21jpmhw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*pvst0qtiOIAJvVWyxcuDgg.png" /></figure><p>The Program must always encourage their students as well as the faculty to think beyond the curriculum. In this regard, the Program must provide additional facilities. These facilities will help the students and the faculty in experimenting in newer domains to feed their hunger for knowledge. The more the Program is dedicated itself to the future of the students, the more such additional facilities will be instantiated. Hence this can be another measure of the intent of the Program for its quality education.</p><p>The overall efficiency of the Program depends on the ambience of the laboratories, classrooms. Proper maintenance and cleanliness will surly improve the work efficiency. The ambience and the maintenence of the Program hence is another measure of intent towards excellence.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G96aoqILcVkRDXjL_Is_5w.jpeg" /></figure><p>Academic project is the stage wherein all the knowledge gathered in the first three years of engineering is applied together to convert some idea into reality. Here the students realize the importance of each course studied during the Program and gain an insight on how to fuse the concepts from various courses to achieve a target. This activity is not time bound and the work is carried out mostly on instinct. The work in projects can be realized only when we have a dedicated fully equipped facility for the projects available to the students all the time. The students who spend their time for the project work will surely have an upper hand as they learn many things like concept fusion, time management, resource management, team work, leadership etc. The quality of the facilities provided for the project work will be a direct measure of the Program’s attitude towards the students&#39; growth.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*kfsXHopx9GLCHAzDwyxr1Q.jpeg" /></figure><p>Safety measures in the laboratories is the prime concern for any Program. It shows how much the Program value their stakeholders and the equipment. Proper instructions must be made visible at all the important spots that help the people follow during emergencies.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*QiblLGH8o79MH00CfybO3Q.jpeg" /></figure><h4>7 Continuous Improvement</h4><p>The strategy of the program must be too ensure continues improvement in the quality of the Program with time. It must be taken up quantitatively and visualize the continuous growth. This growth can be measured by the academic performance of the students, through the placements- number, type of companies visiting our campus for placements, packages etc., and the through the higher studies information- number, global ranking of the universities, type of Programs etc.</p><p>The Program Outcomes and Program Specific Outcomes okay an important role in measuring this growth in academic performance.</p><p>The whole process begins at the beginning of each semester. At this stage the target of each of the POs and PSOs for each course is declared. At the end of the semester, the results of the students is analysed to find how much of our target is attained. The aim is to achieve at least 90 percent of the target. If any PO or PSO fails in attaining the threshold, then we need to introspect our process and identify the gaps. These gaps must be filled up in the next academic year through various activities. All this analysis is well documented and preserved for the future actions.</p><p>If the attainment is beyond the target, the Program’s quality is underestimated and hence the higher target is planned for the next academic year.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*mFLwWIb-p11gbdixJaskiQ.jpeg" /></figure><p>The better the quality of education, the better will be the students&#39; performance, and hence better companies will visit our campus for placements. The better the performance of our students, the better will be the salary packages offered. The quality of these companies, the number of students placed, and the packages offered will mirror directly how out stakeholders accept our students globally. This will directly assess the Program as well as the institute.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NCAxnkh7J689TK6eQcL8ew.jpeg" /></figure><p>On similar grounds, the better is the education system, the better qualified the students will be and hence can easily get admitted to top ranking universities for their bigger studies. The number of students admitted, the ranking of the universities admitted to, will directly reflect the Program’s quality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/460/1*i8k0lTshs3SmLirLjK0CtQ.jpeg" /></figure><p>With time is the Program continues to improve is performance, the overall ranking of the Program will increase and hence better students will opt the Program in the future. The higher performance of the Program will attract better studentsfor admissions. These better quality students will further improve the overall ambience of the program and hence an avalanche effect in the Program’s quality and hence the growth.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/780/1*4UH9ODLbJh8B8Q_VgP1MVg.jpeg" /></figure><p>The steps 2 through 7 are repeated on regular basis to ensure monotonic improvement in the quality of the Program, institute, the students and hence the society.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c3bc3538d06d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How does Machine Learning influence our lives?]]></title>
            <link>https://medium.com/@svm161265/how-does-machine-learning-influence-our-lives-781bbfd00e43?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/781bbfd00e43</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Fri, 05 May 2023 02:47:25 GMT</pubDate>
            <atom:updated>2023-05-05T02:47:25.161Z</atom:updated>
            <content:encoded><![CDATA[<p>Hi friends,</p><p>Remember the last time you went to departmental stores for one or two items? I bet you came out buying at least five items, you are lucky if this list contains the items for which you got into the stores. What actually happened is your emotions are played with through machine learning.</p><h3>How do they do it?</h3><p>Without our awareness, our spending patterns are being studied. The first thing they ask at the cash counter is our cell numbers. Next we’re encouraged for digital payment. This is the simplest way of collecting our contact information and then to study our spending patterns.</p><p>The data hence collected, along with the items purchased, can be a great source of information if used optimally. It gives the daily, weekly, seasonal sales patterns. The business corporates exploit out emotions through this information.</p><h3>Why machine learning?</h3><p>Understanding the insights of the emotions behind the purchases is not a simple task when the stores is dealing with thousands of items and has thousands of customers visiting daily. The better insights can be made possible when this data is collected over sufficiently long periods. That makes billions of combinations to be analysed and is beyond human capabilities forcing us to make machines to act for.</p><h3>How machine learning can help?</h3><p>The machines are made to learn the insights of the data. The machines are trained to correlate the data. The data contains sales details, inventory details, shelf life of the items. With this data, the machines can understand the trends and seasonality of the sales. It makes impacts on many dimensions:</p><ul><li><strong>Future Sales Predictions</strong>: With the knowledge of the trend and seasonality of the sales, the machine can accurately predict the sales size item wise.</li><li><strong>Inventory control</strong>: The future sales predictions can help in maintaining the required inventory through in time purchases.</li><li><strong>Sales strategies</strong>: With the knowledge of future sales predictions, shelf life of the items, many strategies are possible through machine learning. Let me discuss these in detail in the next section.</li></ul><h3>Sales strategies play over our emotions</h3><p>At the end of each day, the whole team of the stores is busy in rearranging the positions of the items in the shelves. The shelves usually are logically split into the parts height wise. The middle portion is in line of sight with the customer. This portion of the shelf is easily visible and hence attracts the customers immediately.</p><ul><li><strong>Middle portion of the shelves</strong>: This region immediately catches the attention of the customers. Hence is reserved for the items that earn more profit but are not popular brands. As per the famous Hindi saying “<strong><em>jo dikhta hai woh bikhta hai!</em></strong>” meaning “<em>what you see is sold!</em>”, these items, though not popular are simply sold because they’re made visible. Items with lesser shelf lives also occupy this space. This way, the high profit making but non popular items as well as those with less shelf life are easily sold.</li><li><strong>Upper and lower portions of shelves</strong>: This area is meant for popular products, high quality products, items with less profits and having longer shelf lives. The popular products and high quality products find their sales wherever they’re placed. Customers search for them anyway.</li><li><strong>Items bought together</strong>: To ensure that the customer visits each part of the stores, the items that are commonly bought together are placed in different sections of the store. Now, while searching the required products, the customer has to cross many sections. And during this walk, many other items attract the attention of the customers. This will increase the chances of buying unnecessary items.</li><li><strong>Discounts</strong>: The products that have more significance on the day will bear no discount. To encourage the sales with lesser demand, discounts are offered. The lesser the demand, the more will be the discount.</li><li><strong>At the entrance</strong>: To ensure the sale of dead stock, these items are well decorated at the entrance to tempt the customers to buy even if they never need it.</li><li><strong>Ads</strong>: After your multiple visits to the store, your purchase habits are well studied and announcements of discounts, offers etc are sent to your cell number. Most of the times, these offers contain the items you never intended to buy or more discounts for less frequently purchased items and small or no discounts you regularly buy. This way they compel you to buy even unnecessary items along with the necessary ones.</li></ul><p>Each day, the whole trend keeps changing and to have maximum profits, the stores must update all the above-mentioned information on daily basis. The machine learning can be a handy tool for analysis of the data available.</p><p>Through Machine Learning, the whole game plan is prepared on daily basis to ensure all the above-mentioned strategies. Through machine learning, we can identify the importance of each item for the day. This in turn help in fixing its position in the shelves, sections of the store, and amount of discount.</p><p>The machine learning algorithm will study the ocean of data on real time basis and predicts the future sales, suggesting the inventory control and the strategies to increase the profits earned.</p><blockquote>The machine first understands our behaviour and exploit our weaknesses !</blockquote><p>All our activities are under careful observations and machine learning helps them to identify our spending habits that in turn be used to encourage us to spend more.</p><p>Be careful.</p><p>That’s it for now. See you soon with some more such issues.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=781bbfd00e43" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[￼]]></title>
            <link>https://medium.com/@svm161265/-19247b121b24?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/19247b121b24</guid>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Fri, 05 May 2023 02:38:32 GMT</pubDate>
            <atom:updated>2023-05-05T02:38:32.895Z</atom:updated>
            <content:encoded><![CDATA[<p>Humans are always trying to simplify their efforts in doing anything for their living. At the beginning of evolution, the humans had only nails and teeth as tools for their survival. Then they realised that they can make use of the stone chips, sticks and bone pieces for hunting and other activities. Then with time, unsatisfied with the current performance, learnt to develop tools using the branches, shoots and stone chips for long distance targets. This gave the humans an opportunity to think beyond survival.</p><p><a href="https://www.vectorstock.com/royalty-free-vector/primitive-culture-tools-stone-age-stone-or-wooden-vector-35358302">(Source: https://www.vectorstock.com/royalty-free-vector/primitive-culture-tools-stone-age-stone-or-wooden-vector-35358302)</a></p><p>Once the food and safety issues are taken care of, the humans began looking into the ways to do the acts more efficiently. The prime concern now was to reduce the efforts needed to do the regular daily activities like growing food, transportation etc. This led to agriculture and machines. And the humans wants started to rock high. Humans are never satisfied. This quality of being always unsatisfied forced to develop more and more.</p><p><a href="https://en.m.wikipedia.org/wiki/Simple_machine">(Source: https://en.m.wikipedia.org/wiki/Simple_machine)</a></p><p>The advent of electricity further increased this pace of development. With all this, the humans could only reduce the usage of muscle power required to do any task.</p><p><a href="https://www.electricalclassroom.com/electric-machines/">(Source: https://www.electricalclassroom.com/electric-machines/)</a></p><p>Now the human wanted someone or something to think on his behalf. This forced one to create machines like calculators, computing machines. These machines act very fast and compute anything at high speed and more accurately. Human was happy that the mental activities are now possible through machines.</p><p>But, is human satisfied? Never. Now what? Human now started to realise that the machine is useless unless a human codes the machine regarding what to do. This coding took lot of human efforts. This is due to the very fact that the coding had to be done in a language that the machine understand. This language needs to be unambiguous. But there is a lot of ambiguity in human languages. Each sentence of human is context based and the machine need a context free language. This approach is code- centric. Here the code is of primary concern whereas data is secondary. In addition the coding of the real-time problems is almost beyond the human limits. An alternative to coding was the need of the hour.</p><p>One alternative is to make the machine understands the human needs by itself and solve it by itself. All this must be done with no or minimal human interaction. The idea is to give commands to the machine as if giving to an intelligent human. This led to the new phrase &quot;Artificial Intelligence&quot; famous as AI. The first step towards AI is to make machine learn the way humans do. That&#39;s the concept of machine learning. Today enormous data is available. This data can be of great help in solving our problems through machines. AI focuses on transformation from code- centric to data- centric approach.</p><p><a href="https://www.analyticsinsight.net/artificial-intelligence-the-history-now-and-future/">(Source: https://www.analyticsinsight.net/artificial-intelligence-the-history-now-and-future/)</a></p><p>Now, what&#39;s learning?</p><p>Learning is the process of acquiring new understanding, knowledge, behaviors, skills, values, attitudes, and preferences [1].</p><p>This process of learning in humans need to be understood first before trying to reach a machine.</p><p>Now let&#39;s try to go back to our childhood to see how we learnt. But due to the abstraction, we&#39;ve forgotten how we learnt. We remember only what we learnt. That&#39;s the next challenge- understand how we learnt!</p><p>So let&#39;s observe how a child learns. There are mainly three modes of learning in a child&#39;s life.</p><p>* Imitate others</p><p>* Observe the surrounding</p><p>* Trial and error</p><p>In the first place, the child tried to imitate other&#39;s actions- moving limbs, talking, eating etc. The next step is others helping the child to imitate. Now the child tries to understand other&#39;s instructions. This phase of learning, some sort of supervision is required. This phase of learning is called Supervised learning. Here the supervisor or the teacher shows something and tells the child what it is. It can be learning alphabets, for example. In this scenario, the teacher shows the various characters written on the board and reads out how to pronounce it. This process is repeated multiple times till the child is able to correlate the symbol with its pronunciation. What happens in the child&#39;s brain during this process is neither known to the teacher nor the child. The teacher&#39;s work is to give two instances to correlate. Be it alphabet, maths or anything, the process of teaching is the same. This abstraction can be the stepping stone to machine learning.</p><p>From the moment the child is born, it keeps keenly observing the neighborhood. It always tries to differentiate between the objects around. Mother from others, known faces from unknown faces, humans from other living creatures, lubing creatures from nonliving ones etc. In this phase, no one is there to follow or teach. The whole learning is done automatically without anyone&#39;s aid. This is Unsupervised learning. The main features that assist one to differentiate are the color, shape, texture, smell, touch and taste. These features help in clustering the objects. The objects with similar features are in same cluster whereas the objects with dissimilar features belong to different clusters. Most of these features used cannot be expressed quantitatively. They can only be expressed qualitatively. For example, try explaining the fragrance of mango 🥭 🤔. It cannot be expressed in words but can only be experienced.</p><p><a href="https://www.researchgate.net/figure/Supervised-learning-and-unsupervised-learning-Supervised-learning-uses-annotation_fig1_329533120/amp">(Source: https://www.researchgate.net/figure/Supervised-learning-and-unsupervised-learning-Supervised-learning-uses-annotation_fig1_329533120/amp)</a></p><p>In the next stage, the child has mobility and can act on its own. Here the child starts to experiment with the neighborhood. Touching, eating and like. In this process, it experiences good or bad. Now the child continues to repeat the actions that gave good experience and avoid the actions with bad experiences. This is Reinforcement learning. A simple example of reinforcement learning is learning to ride a bicycle🚲. This neither can be learnt observing how others ride nor reading or listening to others experiences. The rule for riding is simple:</p><p>- keep the bike moving by peddling</p><p>- maintain balance by proper handle oscillation</p><p>If any of these missed- crash😭. Just knowing these rules is not sufficient to ride. We need to ride, fall, ride again, fall again.... After each fall if we understand what actions are good and what are bad, we then encourage ourselves to use good actions more often that the ones that are bad. This is Reinforcement learning.</p><p><a href="https://www.guru99.com/reinforcement-learning-tutorial.html">(Source: https://www.guru99.com/reinforcement-learning-tutorial.html)</a></p><p>The similar process can be used while making the machine to learn.</p><p>The simplest form is to provide both the input as well as the corresponding output to the machine and let the machine to correlate them. This is the Supervised learning approach. This approach is useful when the data we have contains both the input as well as output. In this method a model is created by training. A model is essentially a mathematical model with many adjustable parameters. During training, the input is applied to the model and it&#39;s output if compared with the required output. The error between the actual output required and the one getting from the model is minimized by adjusting the parameter values. Once the error is within tolerance limit for all the input/ output pairs, the model is ready.</p><p>There are two types of supervised learning depending upon the type of output. If the output is continuous, then the learning is called Regression. This method correlate the output with the input through mathematical expressions. Once this model is ready, we can supply any input to the model and expect a well approximated output. In the other hand, if the output is categorical, then the learning is Classification. This method tries to partition the domain into sub-domains such that each sub-domain contains one category of data.</p><p>(<a href="https://dev.to/petercour/machine-learning-classification-vs-regression-1gn">https://dev.to/petercour/machine-learning-classification-vs-regression-1gn</a>)</p><p>The next form of machine learning is Unsupervised learning. In this category, the data contains only the input and no output is supplied. One way is to look at these features and cluster the data into various clusters using similarity and dissimilarity measures. This process of Unsupervised learning is called Clustering. The data with similar features are clustered together. The similarity measure cab be decided either by number of clusters or by fixing the maximum size of each cluster.</p><p>(<a href="https://www.researchgate.net/figure/Clustering-example-with-intra-and-inter-clustering-illustrations_fig1_344590665/amp">https://www.researchgate.net/figure/Clustering-example-with-intra-and-inter-clustering-illustrations_fig1_344590665/amp</a>)</p><p>Other application of Unsupervised learning is to reduce the complexity of the problem by simplifying the data set. This can be done by reducing the dimensionality of the data. This can be achieved by eliminating redundant rows and columns in the data or by extracting the features smd transforming into a new simplified domain. This simplification had many advantages. The data is cleaned or of redundancy and compact in size and hence working with this data is simple and faster. This process is called Dimensionality reduction.</p><p>(<a href="https://www.pinecone.io/learn/dimensionality-reduction/">Source:https://www.pinecone.io/learn/dimensionality-reduction/</a>)</p><p>Then comes the most exciting form of learning. In this case that data is created while learning. The whole game begins with some rules. The system contains an environment upon which an agent acts. These actions produce either positive or negative results. The agent exploits the environment by repeating the actions producing positive results more often and avoiding those produced negative results. With time, there is monotonous improvement in agent&#39;s efficiency. This is the ultimate form of learning and is called Reinforced learning.</p><p><a href="https://enterrasolutions.com/is-reinforcement-learning-the-future-of-artificial-intelligence/">(Source: https://enterrasolutions.com/is-reinforcement-learning-the-future-of-artificial-intelligence/)</a></p><p>Lets continue our detailed discussion in the future sessions.</p><p>See you soon 😊</p><p>1. https://en.m.wikipedia.org/wiki/Learning</p><p>-Dr. Sandeep V.M. 💡💻🔆🪔</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=19247b121b24" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Humans are always trying to simplify their efforts in doing anything for their living.]]></title>
            <link>https://medium.com/@svm161265/humans-are-always-trying-to-simplify-their-efforts-in-doing-anything-for-their-living-a4871b9e0f62?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/a4871b9e0f62</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Fri, 05 May 2023 02:37:38 GMT</pubDate>
            <atom:updated>2023-05-05T02:37:38.588Z</atom:updated>
            <content:encoded><![CDATA[<p>Humans are always trying to simplify their efforts in doing anything for their living. At the beginning of evolution, the humans had only nails and teeth as tools for their survival. Then they realised that they can make use of the stone chips, sticks and bone pieces for hunting and other activities. Then with time, unsatisfied with the current performance, learnt to develop tools using the branches, shoots and stone chips for long distance targets. This gave the humans an opportunity to think beyond survival.</p><p>Once the food and safety issues are taken care of, the humans began looking into the ways to do the acts more efficiently. The prime concern now was to reduce the efforts needed to do the regular daily activities like growing food, transportation etc. This led to agriculture and machines. And the humans wants started to rock high. Humans are never satisfied. This quality of being always unsatisfied forced to develop more and more.</p><p>The advent of electricity further increased this pace of development. With all this, the humans could only reduce the usage of muscle power required to do any task.</p><p>Now the human wanted someone or something to think on his behalf. This forced one to create machines like calculators, computing machines. These machines act very fast and compute anything at high speed and more accurately. Human was happy that the mental activities are now possible through machines.</p><p>But, is human satisfied? Never. Now what? Human now started to realise that the machine is useless unless a human codes the machine regarding what to do. This coding took lot of human efforts. This is due to the very fact that the coding had to be done in a language that the machine understand. This language needs to be unambiguous. But there is a lot of ambiguity in human languages. Each sentence of human is context based and the machine need a context free language. This approach is code- centric. Here the code is of primary concern whereas data is secondary. In addition the coding of the real-time problems is almost beyond the human limits. An alternative to coding was the need of the hour. <br>One alternative is to make the machine understands the human needs by itself and solve it by itself. All this must be done with no or minimal human interaction. The idea is to give commands to the machine as if giving to an intelligent human. This led to the new phrase &quot;Artificial Intelligence&quot; famous as AI. The first step towards AI is to make machine learn the way humans do. That&#39;s the concept of machine learning. Today enormous data is available. This data can be of great help in solving our problems through machines. AI focuses on transformation from code- centric to data- centric approach.</p><p>Now, what&#39;s learning? <br>Learning is the process of acquiring new understanding, knowledge, behaviors, skills, values, attitudes, and preferences [1]. <br>This process of learning in humans need to be understood first before trying to reach a machine. <br>Now let&#39;s try to go back to our childhood to see how we learnt. But due to the abstraction, we&#39;ve forgotten how we learnt. We remember only what we learnt. That&#39;s the next challenge- understand how we learnt! <br>So let&#39;s observe how a child learns. There are mainly three modes of learning in a child&#39;s life.<br>* Imitate others<br>* Observe the surrounding<br>* Trial and error<br>In the first place, the child tried to imitate other&#39;s actions- moving limbs, talking, eating etc. The next step is others helping the child to imitate. Now the child tries to understand other&#39;s instructions. This phase of learning, some sort of supervision is required. This phase of learning is called Supervised learning. Here the supervisor or the teacher shows something and tells the child what it is. It can be learning alphabets, for example. In this scenario, the teacher shows the various characters written on the board and reads out how to pronounce it. This process is repeated multiple times till the child is able to correlate the symbol with its pronunciation. What happens in the child&#39;s brain during this process is neither known to the teacher nor the child. The teacher&#39;s work is to give two instances to correlate. Be it alphabet, maths or anything, the process of teaching is the same. This abstraction can be the stepping stone to machine learning. <br>From the moment the child is born, it keeps keenly observing the neighborhood. It always tries to differentiate between the objects around. Mother from others, known faces from unknown faces, humans from other living creatures, lubing creatures from nonliving ones etc. In this phase, no one is there to follow or teach. The whole learning is done automatically without anyone&#39;s aid. This is Unsupervised learning. The main features that assist one to differentiate are the color, shape, texture, smell, touch and taste. These features help in clustering the objects. The objects with similar features are in same cluster whereas the objects with dissimilar features belong to different clusters. Most of these features used cannot be expressed quantitatively. They can only be expressed qualitatively. For example, try explaining the fragrance of mango 🥭 🤔. It cannot be expressed in words but can only be experienced.</p><p>In the next stage, the child has mobility and can act on its own. Here the child starts to experiment with the neighborhood. Touching, eating and like. In this process, it experiences good or bad. Now the child continues to repeat the actions that gave good experience and avoid the actions with bad experiences. This is Reinforcement learning. A simple example of reinforcement learning is learning to ride a bicycle🚲. This neither can be learnt observing how others ride nor reading or listening to others experiences. The rule for riding is simple: <br>- keep the bike moving by peddling<br>- maintain balance by proper handle oscillation<br>If any of these missed- crash😭. Just knowing these rules is not sufficient to ride. We need to ride, fall, ride again, fall again.... After each fall if we understand what actions are good and what are bad, we then encourage ourselves to use good actions more often that the ones that are bad. This is Reinforcement learning.</p><p>The similar process can be used while making the machine to learn. <br>The simplest form is to provide both the input as well as the corresponding output to the machine and let the machine to correlate them. This is the Supervised learning approach. This approach is useful when the data we have contains both the input as well as output. In this method a model is created by training. A model is essentially a mathematical model with many adjustable parameters. During training, the input is applied to the model and it&#39;s output if compared with the required output. The error between the actual output required and the one getting from the model is minimized by adjusting the parameter values. Once the error is within tolerance limit for all the input/ output pairs, the model is ready.</p><p>There are two types of supervised learning depending upon the type of output. If the output is continuous, then the learning is called Regression. This method correlate the output with the input through mathematical expressions. Once this model is ready, we can supply any input to the model and expect a well approximated output. In the other hand, if the output is categorical, then the learning is Classification. This method tries to partition the domain into sub-domains such that each sub-domain contains one category of data.</p><p>The next form of machine learning is Unsupervised learning. In this category, the data contains only the input and no output is supplied. One way is to look at these features and cluster the data into various clusters using similarity and dissimilarity measures. This process of Unsupervised learning is called Clustering. The data with similar features are clustered together. The similarity measure cab be decided either by number of clusters or by fixing the maximum size of each cluster. <br>Other application of Unsupervised learning is to reduce the complexity of the problem by simplifying the data set. This can be done by reducing the dimensionality of the data. This can be achieved by eliminating redundant rows and columns in the data or by extracting the features smd transforming into a new simplified domain. This simplification had many advantages. The data is cleaned or of redundancy and compact in size and hence working with this data is simple and faster. This process is called Dimensionality reduction. <br>Then comes the most exciting form of learning. In this case that data is created while learning. The whole game begins with some rules. The system contains an environment upon which an agent acts. These actions produce either positive or negative results. The agent exploits the environment by repeating the actions producing positive results more often and avoiding those produced negative results. With time, there is monotonous improvement in agent’s efficiency. This is the ultimate form of learning and is called Reinforced learning.<br> Lets continue our detailed discussion in the future sessions. <br>See you soon 😊</p><p>1.https://en.m.wikipedia.org/wiki/Learning</p><p>-Dr. Sandeep V.M. 💡💻🔆🪔</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a4871b9e0f62" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[An interesting story on signal convolution]]></title>
            <link>https://medium.com/@svm161265/an-interesting-story-on-signal-convolution-47d11765bbd8?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/47d11765bbd8</guid>
            <category><![CDATA[timeseries]]></category>
            <category><![CDATA[emotion-control]]></category>
            <category><![CDATA[convolution]]></category>
            <category><![CDATA[emotions]]></category>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Fri, 08 Oct 2021 13:53:34 GMT</pubDate>
            <atom:updated>2021-10-08T13:53:34.540Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/202/1*ENbqj5c98SibAD_lvQnnXg.png" /></figure><p>One day a friend of mine looked very confused. When I asked the reason, he said that the convolution is the problem. He went on narrating his problems:</p><ul><li>What is convolution?</li><li>Why is it given so much importance in signal processing?</li><li>Why the time series of the window function is reversed while performing the convolution?</li><li>And many more…</li></ul><p>Then I said let me clarify all your doubts through a simple example.</p><blockquote>Now tell me what issues control your mood? Surely your past experiences will decide your mood at this moment. Isn’t it true?</blockquote><blockquote>You say — Yes, definitely.</blockquote><p>My next question to you is:</p><blockquote>Do you totally remember everything from the past?</blockquote><blockquote>You say no.</blockquote><blockquote>Yes, we can’t remember everything. Our brain has some mechanism to forget what we know with time, unless it’s renewed time to time. So under normal conditions, the intensity with which we remember keeps decreasing with time.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/373/1*FbFOsnbhZhf9i-ArdOChLg.jpeg" /></figure><blockquote>Figure 1: Recollection of an event vs time</blockquote><blockquote>We keep forgetting and our memory about the past becomes weaker with time. In other words, whatever happens with us now, we remember fully. But a day later, we tend to have lesser influence. The next day even lesser and so on as shown in figure 1.</blockquote><blockquote>But the today’s emotions not only depend on what happened today but also in the past. Hence it’s a cumulative effect of all the past experiences that we can remember and each of the past experience has different scale of remembrance depending on the delay.</blockquote><blockquote>As of now, your mood depends upon what’s 100 percent of today’s memory of events, plus maybe 90 percent of yesterday’s, plus 80 percent of day before yesterday’s and so on. This total effect created your mood on today.</blockquote><p>So we’re taking of two time series</p><ul><li>The capacity of remembering an event with time.</li><li>Experiences we had in the past.</li></ul><p>Figure 1 gives the amount of experience we remember about an event after the given number of days. But this gets us to <strong>future</strong>. But we are <strong>not</strong> working in future.</p><p>What we need to know is how much we can recollect about our past experiences today at this instant. This means we’ve to go back into past and check how much we can recollect. We can remember everything about the present experience, but maybe only 90 percent of yesterday’s, 80 percent of day before yesterday’s and so on. <strong>This forces us to reverse the time in the time series</strong> as shown in figure 1. This leads us to figure 2.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/373/1*Sb4VItPtdjqcyf7UMLeEMw.jpeg" /></figure><blockquote>Figure 2: Reversed time-series — Recollection of past as of now vs time.</blockquote><p>Hence at any given instance, the mood now depends upon all the past experiences with coefficients as shown in figure 2.</p><p>The process can be summarised as an expression shown in equation 1.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/383/1*wfu_LLvncL8cv-FnyKwfSg.png" /></figure><blockquote>Equation 1.</blockquote><p>Here x is our experiences and h is our capacity of recollect the past.</p><p>The quantity <strong>h is defined for future</strong> — how much of today we can remember in the future. But what <strong>we need is of the past</strong>. That is, how much of the past we can recollect now. Hence <strong>this time series of the coefficients is reversed</strong> for the actual usage.</p><p>In order to find the emotions of a person over a period of time, then we need to repeat this process by shifting the time series h such that the present of the h is made to coincide with the day of experience for which day you want to know the emotions.</p><blockquote>Now let me summarise what the convolution process is:</blockquote><blockquote>It’s just sliding the reversed time series h, known as window, over the actual signal. At each position of h</blockquote><ul><li>Compute the product of the coefficients given by the reversed h and the corresponding signals overlapping</li><li>Sum all these products</li></ul><p>This sum is the convolution response at that instant of time of the signal for which the zero reference of h coincides.</p><p>In short, the convolution is a process of finding the cumulative effect of the signal over a certain period — decided by the length of the time series h and the effect type is decided by the shape of h.</p><p>For example, if h is made up of 10 values equal to 0.1 each positioned at 0,1,2,….9, then the convolution of this h with any signal will be the average of last 10 samples of x at any given time t.</p><p>The values shown in figure 1 are for some common personality. For some other personality, the shape and size of h will be different.</p><blockquote>In our example, h represents the behaviour of a person, x is the various situations the person is forced to face and the convolution of these two is the response of that person to the situations faced.</blockquote><blockquote>For some special personality who lives only in the present and whose decisions are unaffected by the past experiences, then for such special person, h has a 1 at t=0 and 0 elsewhere in the range. Such a h is called an impulse and the convolution of any signal x with an impulse is x itself — called impulse response.</blockquote><blockquote>The impulse response serves a great job of identifying the behaviour of the unknown system. Just convolve the impulse input with the system and the output we get is itself the system. Surely you’re confused with what this system is? Yes, let me get details of this in my next article.</blockquote><h3>Hope with my friend, your doubts about the convolution is cleared.</h3><p>Many more such issues in the queue.</p><p>Keep waiting….</p><blockquote>Not to forget 👏👏👏 if you found some interesting concepts clarified in this article and please share with your friends.</blockquote><p>Thanks.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=47d11765bbd8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When and why to use P, PI, PD and PID Controller?]]></title>
            <link>https://medium.com/@svm161265/when-and-why-to-use-p-pi-pd-and-pid-controller-73729a708bb5?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/73729a708bb5</guid>
            <category><![CDATA[phase]]></category>
            <category><![CDATA[pd]]></category>
            <category><![CDATA[pi]]></category>
            <category><![CDATA[pid-controller]]></category>
            <category><![CDATA[controller]]></category>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Wed, 06 Oct 2021 04:49:32 GMT</pubDate>
            <atom:updated>2021-10-06T04:49:32.565Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/680/1*DAA0DeTyQH8DCNGyCFCTww.png" /></figure><p>When the plant’s performance isn’t as per the system’s requirements, a closed loop control system can be handy, instead of a new plant. The error between the actual and the required outputs derived the controller. This controller in turn produces the necessary correcting actions upon the plant pulling the actual output towards the required one.</p><p>The easiest approach is to produce the actuating signal u proportional to the error, the more is the error, the higher will be the actuating signal magnitude. This allows the actual output to follow the required output. This is the <strong>Proportional controller.</strong> This controller is useful when the error is under some limit. If the error is beyond some threshold, just a proportional controller is not sufficient to handle the error.</p><p>There are two types of errors:</p><ul><li>Transient error</li><li>Steady state error</li></ul><p>When the input is applied, there is some delay in the reaction of the plant to the input due to the reluctance. If this delay is tried to reduce by increasing the gain of the <em>Proportional</em> controller, it overshoots and oscillations at the output are produced. This is the <em>transient error.</em> Now it’s time to look at the rate of change of error due the control actions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/1*dNVaLc1s-_y6TRpegCoZ_A.png" /></figure><p>The rate of change of error at the time of crossing the final value is maximum and if this is made to control the output of the plant, then the overshoot as well as the oscillatory behaviour can be put under control. This is the idea being the <strong>derivative controller</strong>. This controller should be used very carefully when the signal contains noise, which if very common. The derivative controller amplifies the noise and throws the plant into instability.</p><p>On the other hand, plant may never reach the set point due to some inherent limitations. This leads to the steady state error. The integral of the error will be a better solution to identify this steady state error. The integral of the error can then be used to minimise this steady state error. This idea of using the constant error over time is the <strong>Integral controller</strong>. This controller improves the performance of the plant under noisy environments.</p><h3>PD Controller</h3><p>Practically, no single controller can serve the purpose. It needs multiple controllers work together to meet the requirements. Whenever, it’s needed to speed up the transient period, we add proportional control and the overshoot and the oscillatory problems need the derivative control. This calls for <strong>Proportional- Derivative controller(PD Controller)</strong>. Special care must be taken to ensure that the signals are noiseless, else the noise can get amplified by the derivative actions and creates stability problems to the plant.</p><h3>PI Controller</h3><p>This helps in the situation where proportional control is necessary to speed up the settling and integral control to reduce the error that is constant over time. This is <strong>Proportional Integral controller (PI controller)</strong>.</p><h3>PID Controller</h3><p>More general case would be to address all the above discussed problems with a single solution. In this case all the three controllers are used in parallel will appropriately chosen gains. Adjusting these gains, any combination of P,I,D controllers can be obtained and hence a more robust one.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/685/1*TjnoY8-jW4rfEdpvinXCDA.png" /></figure><p>The three gains — Kp, Ki and Kd for the Proportional, Integral and Derivative controller actions can decide the contribution of each one in the final controller action.</p><h3>When to use</h3><ul><li>A Proportional controller is used to reduce the rise time and speed up the response. This controller makes no changes in the phase response of the plant.</li><li>A Derivative controller is required to minimise the transient errors like overshoot and oscillations in the output of the plant. But this can create heavy instability in noisy environments. Be careful to use smaller gain with this controller. It provides a phase lead to the output when compared with the input. , usually with no change in magnitude.</li><li>An Integral controller corrects the time invariant errors. This provides a phase lag and no change in magnitude in the output.</li><li>A PI controller helps in reducing both the rise time and the steady state errors of the system. To be useful whenever you need to change magnitude and lag the phase together.</li><li>A PD controller reduces the transients like rise time, overshoot, and oscillations in the output. Useful for changing magnitude and want to add phase lead to the output.</li><li>A PID controller is a general form of controller. The gains of the three control actions can be adjusted to achieve any controller. The change in magnitude along with either lead or lag in phase in the output can be made available through this general model of controller.</li></ul><p>To summarise:</p><h3>P:</h3><p><strong>Proportional</strong> controller to reduce the transient period.</p><blockquote>Changes the magnitude only.</blockquote><h3>I:</h3><p><strong>Integral</strong> controller to reduce the time invariant error</p><blockquote>Lags the output phase.</blockquote><h3>D:</h3><p><strong>Derivative</strong> controller to minimize the transient errors like overshoot, oscillatory response.</p><blockquote>Leads the output phase.</blockquote><h3>PI:</h3><p>Reduces rise time and steady state errors.</p><blockquote>Changes the magnitude as well as lags the output.</blockquote><h3>PD:</h3><p>Reduces rise time and transient errors such as overshoot, oscillations in output.</p><blockquote>Changes both the magnitude as well as adds a leading phase to the output.</blockquote><h3>PID:</h3><p>General case of a controller. Can be used to control the magnitude and lead/ lag phase problems.</p><blockquote>Changes the magnitude and can add positive or negative phase to the output as per the requirements.</blockquote><p>That’s it for now.</p><blockquote>Hope you enjoyed this journey into the world of control engineering. If so, please encourage through your 👏 and sharing with your friends.</blockquote><p>Thanks and see you soon with some more stories.</p><p>Bye…</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=73729a708bb5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Machine Learning can simplify the Control System Design?]]></title>
            <link>https://medium.com/@svm161265/how-machine-learning-can-simplify-the-control-system-design-4175adc5d28b?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/4175adc5d28b</guid>
            <category><![CDATA[data-driven]]></category>
            <category><![CDATA[control-system]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[control]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Sun, 03 Oct 2021 11:57:17 GMT</pubDate>
            <atom:updated>2021-10-03T11:57:17.218Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/544/1*dLS3nM-L_HqsE_kPYbn2Lw.gif" /></figure><p>Controlling the system to operate at its optimum is one of the major priorities of any design engineer. It’s also equally important to ensure that the system behaves as per the requirements. The system available usually don’t strictly meet the requirements. Usually it’s preferred to go for closed loop system instead of creating a new system that would fully agree with the requirements.</p><h3>Closed loop Control System</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*UwvmIxLsD0U0-nbm2s1QqA.jpeg" /></figure><p>When the system (called plant here) isn’t behaving as per the requirements, then the output of the plant is continously tracked and any deviation from the expected output will be fed back to create an error signal. The error signal in turn creates the actuating signal through the controller. This actuating signal will ensure that the plant behaves the way it’s intended to.</p><blockquote>The major problem in this design is the need to analytically know the plant in its complete detail. Through time domain analysis, first order or second order behaviors of the plant can be measured. But in reality, the plant will be of higher order. Approximating the plant to a lower order in itself is the first step in degrading the performance. The next source of error is, the design of the controller is carried out for this approximated model of the plant and applied on the actual plant. This leads to imperfect design and further degradation of the performance of the overall system. Then what’s the solution?</blockquote><blockquote>Consider the example of driving the car. Here,</blockquote><blockquote>the car is the plant,</blockquote><blockquote>you are the controller,</blockquote><blockquote>your legs produce the actuating signals such as pressure on accelerator and brake pads,</blockquote><blockquote>the eyes that sense the speed and position of the car as the feedback elements,</blockquote><blockquote>set speed is the input,</blockquote><blockquote>difference between the set and the actual speeds is the error signal,</blockquote><blockquote>this error signal help you in controlling the actuators to maintain the speed of the car.</blockquote><blockquote>In this example, an experienced driver can maintain the speed very easily. But how did he learn to drive? Through experience. Then what’s experience?</blockquote><p>In this driving case, the driver doesn’t know the analytical form of the car dynamics and hence there’s no way to design the controller. But still the driver learns. Here the process of learning is to begin with random assumptions about driving and start working. The brain of the driver records all the actions of the driver and the reactions of the car. Then the brain encourages the driver to take more of the actions that produced positive reactions at the same time discourages for the negative reactions. This process of gathering the experiences is called learning and more specifically, <em>Reinforcement learning</em>.</p><h3>Machine learning</h3><p>Now let’s conduct experiments on the plant and for different inputs, the outputs are measured and documented. Instead of an analytical model of the plant, the machine learning technique is applied to train the machine how the inputs to the plant are mapped to the corresponding outputs. This the knowledge about the plant is created for future use.</p><p>Once the mapping from input to the output of the plant is ready, we can plan to train our machine to inverse map so that our machine now knows what should be fed to the plant in order to achieve a required output.</p><p>The whole process is now data driven and hence no need to explicitly design the controller to achieve the requirements. The controller can be created on self-learning style. The designer’s job here is to provide the I/O data of the plant collected over time and the required behaviour of the plant. The machine learning algorithms allow the machine to learn the control actions needed at different conditions of the plant. A minor retraining of the contort is needed if the plant is modified or the requirements are changed.</p><blockquote>The controller here is trained to have intelligence so as to control the actuators to the plant and achieve the required results. This is similar to gaining the driving knowledge like to press accelerator in order to increase the speed and release it to reduce the speed. In case to suddenly reduce the speed, brakes need to be pressed. The knowledge was gathered while learning: if you press the accelerator the car speeds up and releasing it allows down the car. Whenever the brake was applied, the car speed reduced suddenly.</blockquote><blockquote>We learnt what happens to the car speed when accelerator or brakes are pressed. This knowledge helps the us in future to know what actions to take to drive the car.</blockquote><p>The advantage of data driven controllers is that, the same controller can be used for applications just by retraining it with the new set of data. The change in data will change the controllers behaviors and this in turn will help us in being any requirements.</p><p>To summarise:</p><ul><li>The machine learning helps in creating a dynamic controller.</li><li>No analytical model definition of the plant is required.</li><li>Just the I/O data set is required to train the controller.</li><li>The controller thus designed is dynamic, i.e. the controller can be reprogrammed as and when required.</li><li>Can keep updating the controller behaviour with changes in the plant.</li></ul><p>The machine learning can be a new hope to achieve all those from a given plant that were like dreams due to the non linearity of the plant and limitations in designing the appropriate controllers.</p><p>Enjoy applying machine learning to control systems and achieve higher efficiencies with minimal efforts.</p><p>Hope you enjoyed the article. If so, please👏. And share with your friends.</p><p>Will be back with another issue to discuss.</p><p>Keep waiting…</p><p>Thank you….</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4175adc5d28b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Time domain vs Frequency domain Analysis: What, When and Why’s]]></title>
            <link>https://medium.com/@svm161265/time-domain-vs-frequency-domain-analysis-what-when-and-whys-1e8f91fc3043?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/1e8f91fc3043</guid>
            <category><![CDATA[time-domain-analysis]]></category>
            <category><![CDATA[closed-loop]]></category>
            <category><![CDATA[frequency-domain-analysis]]></category>
            <category><![CDATA[controller]]></category>
            <category><![CDATA[control]]></category>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Fri, 01 Oct 2021 14:49:51 GMT</pubDate>
            <atom:updated>2021-10-01T14:49:51.058Z</atom:updated>
            <content:encoded><![CDATA[<p>For a control engineer, these terms are very frequent. There’s always confusion regarding what, when and why the system should be subjected to time domain or frequency domain analysis.</p><p>Let’s briefly look at what a system analysis is. It’s a process of defining the working or the performance of a system. There are two methods of analysing a system: Qualitative and Quantitative.</p><p>The qualitative analysis is a descriptive explanation of the performance of the system. It’s very impossible for the qualitative analysis be unambiguous. It’s very much difficult when you try to compare systems qualitatively.</p><p>On the other side, it’s very easy to compare the systems quantitatively. However, identifying the parameters that define the systems quantitatively is very difficult issue. But once these parameters are correctly identified, then it’s self explanatory and comparing systems is quite simple.</p><p>Let’s look at some examples to clarify these points. Assume that you’re asked to compare two cars before deciding to purchase. You take test drives and then try to explain your experiences. Qualitatively, you can say the car’s pick up is good. Its smooth while driving. The speed is good. And so on. But these words cannot convince others about the experiences of the narrator.</p><p>Instead, if the experiences are measured like</p><ul><li>the time it takes to reach a speed of 100kmph from rest, or</li><li>the maximum speed of the car</li></ul><p>it is quite easy to compare the performance of the two cars. That’s the power of quantitative data.</p><p>There are two types of quantitative analysis of the system depending upon your response time:</p><ul><li>Time domain</li><li>Frequency domain</li></ul><h3>What is Time domain analysis?</h3><p>Time domain analysis is preferred when the system varies slowly when compared to your response time. This allows you to measure the parameters from time to time. This time-series of the data measured will help us to understand the system behaviour.</p><blockquote>e.g. the time the car takes to pick up to 100kmph speed from rest is the time domain analysis. Our response time is in milli seconds whereas the experimentation time is about 10-15 seconds. We can track the change in speed and record for further analysis.</blockquote><blockquote>Imagine you’re watching a train move from some distance. You can keep noting its position at different instances of time and record for analysis. The train takes minutes to cross from your vision and noting its position is matter of seconds.</blockquote><blockquote>But the same is not possible to track the position of a point in the wheel of the train. Hence time domain analysis can’t be used for such cases.</blockquote><h3>What is frequency domain analysis?</h3><p>When the system is faster than your response time, and if the periodic experimentation can be performed, then frequency domain analysis is preferred. Here the experiments over different frequencies are performed and then analysed.</p><blockquote>e.g. the experiments conducted to measure the mileage of the car while running with different speeds is a frequency domain analysis. Here the car is run for a fixed distance at different speeds and each time the fuel consumed is measured and thereby the mileage. The mileage for different speeds are recorded and analysed.</blockquote><blockquote>Study of rotation of moon (time per revolution) is a time domain analysis while the rotation of wheel of a car (revolutions per unit time) is a frequency domain analysis.</blockquote><h3>Time domain analysis of a system:</h3><p>In this case, a known form of input is applied to the system and the system response is measured for analysing. The simplest form of input is just to start the system working. Mathematically, it’s called step function. Before t=0, the system was off and from t=0 onwards, the system is on. Now the parameters, like speed, of the system are measured as they change after switching on. This change is quantized by appropriate parameters as shown is the figure below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/1*dNVaLc1s-_y6TRpegCoZ_A.png" /></figure><ul><li>Initial value</li><li>Final value</li><li>Rise time</li><li>Peak time</li><li>Maximum overshoot</li><li>Settling time</li></ul><p>The system can be represented without any ambiguity by these parameters listed above.</p><blockquote>Assume that you’re driving a car and aiming to maintain a speed of 80kmph. Depending on your driving style there are three possibilities:</blockquote><ul><li><strong>Arrogant driver</strong>: In this case, the car is suddenly accelerated and before the driver realises, the speed has crossed the limit and hence he may suddenly shows down. It may now go below the set speed which makes him to accelerate a little and cross the limit again. This up and down speeding continues till the steady speed is reached. This is the case of <strong>underdamped</strong> condition (damping factor&lt;1). In this case, both the rise and settling times are lowest. In the negative side, it produces overshoot and oscillatory behaviour.</li><li><strong>Very careful driver</strong>: You slowly accelerate the car and makes sure not to cross the speed limit. In this situating, the car takes a long time to reach almost the set speed but never reaches the exact speed. This is called <strong>overdamped</strong> condition (damping factor&gt;1). No overshoot, no oscillations but very high rise time and almost infinite settling time.</li><li><strong>Perfect driver</strong>: You know the art of driving perfectly and make sure to use appropriate acceleration from time to time and ensure quickly reaching the set speed and maintaining the speed constant there. This is <strong>critically damped</strong> condition (dampingfactor=1). No overshoot, moderate rise and settling times.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/675/1*X61nH8GSjLhgVXc9f8UQ_A.png" /></figure><p>The system representation can further be simplified by knowing the damping condition.</p><h3>System control in time domain</h3><p>When the step input is applied to the system, then ideally a step output is expected. Zero overshoot, zero rise time and zero settling time. But none of the damping value of the system can attain this. The optimal system must have least possible rise and settling times and zero overshoot. This can be made possible by using an underdamped system with closed loop and a proper controller. The underdamped system has very low rise and settling times but has overshoot. This overshoot must be minimized through the controller.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/680/1*DAA0DeTyQH8DCNGyCFCTww.png" /></figure><h3>Controller</h3><p>There are three types of basic controllers.</p><ul><li>Proportional(P): This will vary the magnitude of the output depending upon the error — the difference between the actual output and the intended output.</li><li>Differential(D): In this case, the rate of change of error will control the output. This is useful whenever there is sudden changes in the error. This limits the sudden changes in the error.</li><li>Integral(I): This will vary the output such that more stability is attained earlier. Here the average of the error over a period will control the output.</li></ul><p>Depending upon the requirements, the controller can be any of the above-mentioned controllers or their combinations like PI,PD, PID. Usually combined controllers perform better for any physical system.</p><h3>Frequency domain analysis of a system:</h3><p>When the system behaviour is <em>harmonic</em> and the <em>period is very small compared to our response time</em>, then frequency domain analysis is the only option to study the system. In this set-up, the system is subjected to sinusoidal input of different frequencies and the system response such as the <em>magnitude</em> and the <em>phase</em> of the output is measured. The variation of the behaviour of the system to different frequencies can be studied to understand the system and these measurements can further be used in comparing the performance of two systems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/233/1*tYJU0VwXd4-IMduExOp2Qg.gif" /></figure><p>The above plot shows the mileage of the car at different speeds. This experiment was conducted by running a car with some fixed amount of fuel at different speeds. It can be seen that the car runs efficiently between 30 to 60 mph. This experiment will be helpful in designing the cars. The controlling mechanisms in the car can be trained to improve this efficiency for any given range of speeds.</p><h3>System control under frequency domain</h3><p>The major problem we encounter with feedback system is of sustained oscillations. This is possible when for some frequency the gain around the loop is unity and the overall phase lag is integral multiples of 360⁰. These two must be simultaneously true for a particular frequency. Usually, the gain will be less than unity for the frequency at which the phase lag is 360⁰. The amount of gain that can be increased at the frequency where the phase lag is 360⁰is called the <strong>gain margin</strong>. Similarly, the phase that can be delayed by to make the phase lag of 360⁰ at the frequency where the gain is unity is called the <strong>phase margin</strong>. <em>These two are the better representatives of the system in frequency domain analysis</em>. These two parameters help in understanding the system’s behaviour and useful in comparing the performance of two systems. <em>With negative feedback, the phase lag required is 180⁰ as the other 180⁰ phase delay is provided by the feedback itself.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cnjX2uSn7k08u3JGEQJxUw.png" /></figure><h3>Why such analysis is required?</h3><p>Whenever the system doesn’t satisfy the requirements, then it’s subjected to feedback and controller. The analysis, either in time domain or in frequency domain, performed will help us in identifying the <em>shortcomings</em> of the system as compared to the requirements. This analysis will further provides the means to <em>develop the controller and achieve the required performance</em>.</p><blockquote>Hope you enjoyed reading the article 😀</blockquote><blockquote>Please clap 👏 if you find this article useful and share with your friends.</blockquote><p>See you soon with more articles.</p><h3>Bye…</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1e8f91fc3043" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why encoding required in digital communication?]]></title>
            <link>https://medium.com/@svm161265/why-encoding-required-in-digital-communication-480c83bbdf56?source=rss-9298123e8771------2</link>
            <guid isPermaLink="false">https://medium.com/p/480c83bbdf56</guid>
            <category><![CDATA[digital-communication]]></category>
            <category><![CDATA[psk]]></category>
            <category><![CDATA[ask]]></category>
            <category><![CDATA[fsk]]></category>
            <category><![CDATA[encoding]]></category>
            <dc:creator><![CDATA[Sandeep VM]]></dc:creator>
            <pubDate>Thu, 30 Sep 2021 09:25:07 GMT</pubDate>
            <atom:updated>2021-09-30T09:25:07.089Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZJyljCr3HxSGoO0yFK32FQ.jpeg" /></figure><p>The first operation performed on any digital signal for transmission is encoding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Fu9IQ-mFYsTKF9pL0kQACw.png" /></figure><p>The <strong>infinite bandwidth</strong> of the digital signal compels us to reduce this bandwidth to limited range. Sudden changes in the voltage levels, voltage changes in zero time, make the bandwidth infinite. All the media have a very <strong>limited pass band.</strong> No medium transmits such a signal without distorting. This distorted signal conveys no meanings at the receiving end.</p><p>In case of <em>wireless transmission, the antenna size is related with the bandwidth of the signal being transmitted.</em> It’s size must be half the wavelength of the signal for optimal transmission. Higher the bandwidth, higher will be the variance in the size of antenna required. Practically, the size of antenna is fixed for the mean value of the frequency range. The higher bandwidth, the lesser will be the efficiency. To manage better efficiency, it’s recommended that the bandwidth of the signal is somehow reduced.</p><p><strong>Encoding</strong> comes to our rescue in <strong>reducing the bandwidth</strong> of the signal. The sinusoidal signal is one that’s naturally accepted by any physical system. The sinusoidal signal is characterised by <strong>three parameters</strong> — its <strong>magnitude, frequency and phase.</strong> We can use any of these parameters to carry the information of the digital data either 0 or 1.</p><p>We can vary the <em>Amplitude</em> of the sinusoidal in accordance with the data. We transmit a sinusoidal signal of fixed frequency for a fixed duration if the data is 1, and no signal to transmit a 0.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/1*GaSNOwtQHQyPqn9PPJ6uwA.jpeg" /></figure><p>This is called <strong><em>Amplitude Shift Keying (ASK).</em></strong></p><p>Otherwise, we can use two <em>different frequencies</em> — f¹ for a 1 and f⁰ for a 0. Maintaining these two frequencies closer and yet distinct, we can ensure smaller bandwidth. This is <strong><em>Frequency Shift Keying (FSK).</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/1*avARrmvWVriNhxSW4-hU-Q.png" /></figure><p>The third method is to change the <em>phase</em> of the sinusoidal with the data. In-phase for a 1 and out-of-phase for a 0. This is <strong><em>Phase Shift Keying (PSK)</em></strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/542/1*mwALHQFnjV6cdyQsffx1zg.gif" /></figure><p>That’s the beginning of encoding techniques. Then someone had an idea — if we can change the property of this sinusoidal in four distinct ways — then we can transmit two bits of data simultaneously. Each distinct form for a 00, 01 10 and 11. This is called <strong><em>Quadrature Phase Shift Keying (QPSK).</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/857/1*PB4jP3scH1WgC58TCOb1bg.jpeg" /></figure><p><em>This logic can be extended to include 3,4,5,6 bits together as seen in</em></p><ul><li>8 PSK: 8 distinct phases for 3 bits</li><li>16 PSK: 16 distinct phases for 4 bits</li><li>32 PSK: 32 distinct phases for 5 bits</li><li>64 PSK: 64 distinct phases for 6 bits</li><li>4 PAM: 4 distinct amplitudes for 2 bits</li><li>16 QAM: 16 distinct amplitude-phase combinations for 4 bits</li><li>64 QAM: 64 distinct amplitude-phase combinations for 6 bits</li></ul><p>And many more. These will help us in increasing the data rate of transmission multifold. The data rate with 64 QAM will be 6 times more than that of PSK.</p><p>This encoding with help us in two ways:</p><blockquote>Reduces the bandwidth and hence increases the transmission efficiency</blockquote><blockquote>Allows to speed up the bit rate of data transmission.</blockquote><p>That’s the main idea behind encoding the digital data before transmitting.</p><h3>Hope you enjoyed this article.</h3><h3>Make sure to clap 👏if you like it.</h3><p>Please share with your friends.</p><p>That’s it for now.</p><h3>Will be back soon with more details👍.</h3><h3>Keep waiting🤔…</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=480c83bbdf56" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>