Personal Reflection on Security Engineering
Personal Reflection on Security Engineering
Introduction:
Singer & Friedman (2014) define security architecture as the design and organization of the components, processes, services, and controls appropriate to reduce the security risks associated with a system to an acceptable level, while security engineering is defined as the implementation of that design. The goal of both security architecture and security engineering is first and foremost to protect the confidentiality, integrity, and availability of the systems or business in question, in addition to ensuring other important principles such as privacy (Singer & Friedman, 2014).
As we know, every organization today is highly regulated, and security and risk management are critical for its success. Risk management, risk assessment, threat modeling, and other approaches are used to determine the risks to the system. Once properly identified and assessed, each risk must ultimately be deemed acceptable by the organization, either as is or after an action is taken (Singer & Friedman, 2014).
James Anderson, in 1972, outlined the following fundamentals to any security architecture in the Computer Security Technology Planning Study (USAF):
- Security functions need to be implemented in a manner that prevents their being bypassed, circumvented, or tampered with.
- Security functions need to be involved whenever necessary to implement the security control.
- Security functions need to be as small as possible so that defects are more likely to be found.
Singer & Friedman (2014) talk about ISO/IEC 1924, which the International Organization for Standardization (ISO) published its first revision of standard 19249, “Information technology — Security techniques — Catalog of architectural and design principles for secure products, systems, and applications,” in 2017, where it lists out five architectural principles and five design principles;
5 architectural principles are:
- Domain separation — describes enclosing a group of components together as a common entity.
- Layering — a hierarchical structuring into different levels of abstraction
- Encapsulation — an architectural concept where objects are accessed only through functions that logically separate functions that are abstracted from their underlying object.
- Redundancy — designing a system with replicated components, operating in parallel.
- Virtualization — a form of emulation in which the functionality of one real or simulated device is emulated on a different one.
5 design principles are:
- Least Privilege — access to information should only be granted on an as-needed basis.
- Attack Surface Minimization — the sum of all possible security risk exposure through which an attacker may compromise the system.
- Centralized parameter validation — using a single library to validate similar types of parameters.
- Centralized general security services — centralizing security services can be implemented at several levels.
- Preparing for error and exception handling — ensure that errors are detected and appropriate action taken.
Threat modeling is a process by which potential security threats and vulnerabilities can be identified, and mitigations can be prioritized; three of the most common threat modeling approaches are STRIDE, DREAD, and PASTA (Singer & Friedman, 2014):
- STRIDE helps identify and classify computer security threats. Categories of security threats are spoofing, tampering, repudiation, information disclosure, denial of service, and elevation of privilege.
Get Kiran Limbu Phudong’s stories in your inbox
Join Medium for free to get updates from this writer.
- DREAD model is a quantitative risk analysis model that involves rating the severity of security threats by assigning numeric values (typically from 1 to 10) that represent different levels of severity — key points are: damage, reproducibility, exploitability, affected users, and discoverability.
- PASTA is a seven-step risk-based threat model that supports dynamic analysis — 7 stages are defining objectives, defining technical scope, application decomposition, threat analysis, vulnerability analysis, attack modeling, and risk and impact analysis.
Singer & Friedman (2014) mention some of the important security models, such as the Information Flow Model, Noninterference Model, Bell-LaPadula Model, Biba Integrity Model, Clark-Wilson Model, Brewer-Nash Model, and Take-Grant Model, and they should be selected appropriately. The assessment and mitigation of vulnerabilities of security architectures, designs, and solution elements will also depend on the type of systems, such as Client-Based Systems, Server-Based Systems, Database Systems, Cryptographic Systems, Industrial Control Systems, Cloud-Based Systems, Distributed Systems, Microservices, Computing Systems, Virtualized Systems, etc. (Singer & Friedman, 2014).
Personal Background:
I have had the opportunity to work as a Cybersecurity Analyst for a couple of years in the financial and banking industry, especially in the Identity and Access Management (IAM) and Identity Governance and Administration (IGA) space. As a cybersecurity analyst, I have worked closely in the Access Management area and on a lot of internal controls and compliance. Some of those controls can be listed as Role-Based Access Control (RBAC), Privileged Access Management (PAM), and Lifecycle Event Management. I have also collaborated on various System Development Life Cycle (SDLC) of various applications and catalogs.
Under the Role-Based Access Control, the least privilege access policy will be implemented as the foundational policy. Alongside the least privilege access, other controls such as the Segregation of Duties (SOD) policy are also incorporated. Hence, under RBAC, users will be granted the access necessary for them to perform their day-to-day jobs. Under Lifecycle Event Management, how access will be managed and handled during various lifecycle events, such as joiner (onboarding), mover (transfer), and leaver (termination). It is important to assign the right access to new joiners and transfer, also is equally important to revoke access in case of transfer out and leaver/termination.
Privilege Access refers to all access that can change/update/delete entities, codes, and push codes into production, etc. — also often loosely referred to as elevated access. As part of Access Management, elevated access will need to be requested separately on a need basis, and this access will only last either based on time or project; it will be revoked when no longer required. Service accounts or non-human accounts can also be used to access privileged and elevated access. Access to highly sensitive data and information can also be provided on a token-based and fully monitor of every activity performed within the given login. A clearly defined and documented access management process is key in understanding and utilizing access management, implementing need-based access, and avoiding unauthorized or over-provisioning.
Updating passwords periodically and using unique passwords for each access go a long way in securing our accounts and preventing unauthorized access. Patching and updating software and firmware as part of the device and software/firmware management are mandatory actions performed in a periodic manner. Every employee in the organization is required to go through periodic training regarding data classification and categorizations, compliance training related to data sensitivity — PII and domain-related data such as PCI and HIPAA, etc. Sensitive data and files need to be stored in properly protected drives and devices, which are regularly scanned for non-compliance.
Application of Security Engineering Concepts:
As an RBAC Analyst, the overprivileged access was an issue across and within the organization. Users were easily granted more access than required, which increases the risk of threats or account compromise. This issue was mitigated by redefining the concept and principle of RBAC. Business roles were built on the principle of the least privilege and segregation of duties (SOD). Business roles were also built with a proper approval flow, requiring approval from both the Manager and the Role Owner(s) to approve the access request. The cloud computing environment offers significant advantages, they also poses some serious security challenges, especially in the detection and rectification of unauthorized access (Basavarju, Mahadevachar, Srikantaswamy, 2025). The cloud-aware, model-driven security engineering methodology is proposed to provide a complete approach for designing safe software systems for cloud environments — by integrating security modelling, cloud-specific adaptation, automated transformation to designs and runtime controls, and continuous monitoring, the approach covers important gaps in existing literature and practice (Elengovan, Seshagiri, Sahoo, & Jayabalan, 2026).
Device management is another security issue that can occur at the enterprise level. The lack of updated software and firmware can be a weak link and vulnerable to hacking. In order to mitigate this vulnerability, software and firmware in the devices need to be updated or patched on a regular basis. This enables an improved and stronger firewall.
Password management involves activities of updating user passwords periodically and setting up unique and complex passwords that are difficult to guess by using a combination of upper- and lower-case letters, numeric characters, and special characters. Implementation of password management can help prevent password hacking and, therefore, prevent unauthorized access. Users should not use one password for more than one account. Proper storage of passwords using a password storage device, such as BitLocker, etc., is highly encouraged.
The use of mobile devices can also be very vulnerable, as the use of mobile devices in the enterprise is growing rapidly. The use of mobile devices often means trying to access the company network remotely, which exposes the network further. Some of the mitigation processes for mobile management include the use of multi-factor authentication (MFA), designated software and application downloads, and the use of a secure Wi-Fi network. Encryption is considered the best solution to cloud data security, as it secures data by converting plaintext into ciphertext via encryption and decryption keys (Almuseelem, 2025). Zero trust is another security approach alongside MFA that is used for mobile devices and remote network connections.
The use of AI tools has gained widespread adoption in various sectors, many of which are intertwined with system security engineering practices, especially during the design phase of the software development lifecycle (SDLC (Sharbaf, 2026). The use of AI, however, may have opened a Pandora’s box, as it raises even greater questions. The attacks on the AI model can take various forms, such as tampering with the training data to produce an unbiased model, altering the model’s architecture, or corrupting the model’s information (Sharbaf, 2026). Despite growing interest, substantial knowledge gaps regarding the integration of AI within SSE requirements persist, including limited field experience; therefore, AI should be treated as a security enhancement rather than merely an efficiency gain (Sharbaf, 2026).
Conclusion:
Some of the control mechanisms in overall Access Management are: RBAC, Privilege Access Management, well-defined request, approval, and provisioning processes, etc. MFA and Password management help to keep the authentication process less vulnerable, securing data and the system. Encryption helps with data protection and network protection. Device Management helps protect both devices and the network. The proper use of Cloud-driven models and AI models in designing and implementing security measures can go a long way in boosting security and help overcome shortcomings of the older traditional security measures. Even though AI is helping with automation and cutting costs, it brings up new questions regarding AI-related vulnerabilities and threats, as the existing compliance does not fully incorporate the use of AI and its scope.
Some of the important security models: Information Flow Model, Noninterference Model, Bell-LaPadula Model, Biba Integrity Model, Clark-Wilson Model, Brewer-Nash Model, and Take-Grant Model. Some of the systems are Client-Based Systems, Server-Based Systems, Database Systems, Cryptographic Systems, Industrial Control Systems, Cloud-Based Systems, Distributed Systems, Microservices, Computing Systems, Virtualized Systems, etc. The selection of the model or a combination of models depends on the assessment and mitigation of vulnerabilities of security architectures, designs, and solution elements.
References:
Almuseelem, W. (2025). Enhance the Security of Data Based on Server-Side Encryption in a Cloud Environment. International Journal of Online and Biomedical Engineering. Retrieved on April 20, 2026, from https://research.ebsco.com/c/n37x6l/viewer/pdf/4xrtysywgn
Basavarju, N., Mahadevachar, V.K., & Srikantaswamy, M. (2025). An enhanced secure framework for detecting and rectifying unauthorized access in cloud computing environments using the elliptic curve digital signature algorithm. Engineering, Technology & Applied Science Research, 15 (4), 24188–24195. Retrieved on May 11, 2026, from https://research.ebsco.com/c/n37x6l/viewer/pdf/dexh3l3zer
Elengovan, A.K., Seshagiri, N., Sahoo, I. & Jayabalan, K (2026). Could-aware secure software design using model-driven security engineering. In 2026 International Conference on Electronics and Renewable Systems (ICEARS) (pp. 1226–1231). IEEE. Retrieved on May 11, 2026, from https://ieeexplore.ieee.org/document/11416557
Sharbaf, M. S. (2026). Integrating Artificial Intelligence Into System Security Engineering Design: Enhancing Efficiency And Security. In 2026, IEEE International Systems Conference (SysCon) (pp. 1–6). IEEE. Retrieved on May 11, 2026, from https://ieeexplore.ieee.org/document/11503517
Singer, P.W., & Friedman, A. (2014). Cybersecurity and Cyberwar: What Everyone Needs to Know. Oxford University Press.
Note: MS Copilot was used to verify the accuracy of the APA Citation formats.
