Security and design for real-time systems is becoming more relevant today than ever before. This is mostly due to the technology revolution moving toward automation and optimization, fueled by computer algorithms and software.
In order to cater to these mission-critical functions, whether in aircraft engineering or autonomous vehicles, there is a need to come up with new methods for safe and secure applications. This paper sits very nicely at an intersection of design for optimized allocation, security, and fault tolerance. The authors describe how algorithms can be designed as a directed acyclic graph (DAG) to arrange tasks in the order in which they are meant to happen. This caters to the problem of distributed resources trying to process the application in real time. The nodes and edges are designed to host intermediate nodes, where checksum and recovery can be held. This allows fault tolerance to be introduced in the algorithm, in case this doesn’t go as planned, adding reliability to the design.
The authors have explained the algorithm really well, accompanied by diagrams and tables with examples. They also justify the experiments with multiple optimization techniques and show a comparison of the results in graphical form. These techniques are compared for energy consumption, reliability, and vulnerabilities in the schemes. Their algorithm for reliability and security-constrained optimization does not optimize energy, but security and reliability. Another interesting result is in the discussion on the impact on runtime of tasks. This is especially relevant when designing real-time systems and needing quick responses from applications.
This paper seems very relevant for engineers designing mission-critical artificial intelligence for real-time applications, in any engineering domain. In computer science, normally scientists have represented similar problems using Petri nets, but using DAGs to express the problem also does a good job in this context. This is a recommended read for engineering researchers for real-world applications.
