Emergency Vehicle Priority Systems (A Case Study on Trustworthy CPS)
Smart traffic management is a system wherein centrally-controlled traffic signals and sensors regulate the flow of traffic through the cities in response to specific demand levels. This type of implementation allows cities to:
• Reduce traffic congestion by smoothing traffic flows and prioritizing traffic in response to demand in real time;
• Reduce pollution throughout the city. These systems eliminate stop-start driving, which is inefficient and polluting;
• Give priority to emergency vehicles approaching intersections, phasing lights to give traffic flowing with emergency vehicles a ‘green wave’ through the city;
The Case Study
In an emergency every minute counts. However, in urban areas, ambulances, and other rescue vehicles, often have difficulty making headway due to dense traffic or construction sites. When using sirens and flashing lights, the rescue teams do not have to stop at red lights; however, this can be dangerous for other road users at the intersection. So, what measures can be taken to help fire trucks and ambulances reach their destination more quickly and safely? In this case study, you are getting a task from your client to address the problem. Specifically, you are (as stakeholder) focusing on solutions that enable emergency vehicle priority, namely emergency vehicle priority systems (EVPS).
In this project, modern telecommunication technologies and distributed computing such as e.g., edge computing and LTE are considered in the systems. These technologies allow vehicles to exchange information with other vehicles or the traffic infrastructure, such as traffic lights. AI is also considered in this system to e.g., automate and optimize the decision making. The system prioritizes traffic modes according to different degrees. If a top-priority security or emergency vehicle reports via V2X communication that it needs free passage, the system manages the traffic lights accordingly, allowing the vehicle to pass the intersection without slowing down – while affecting other road users as little as possible.
Some examples of trustworthiness considerations:
EVPS illustrates a highly safety critical system in which it requires real-time performance and reliable error detection. In this case study, there could be potential trade-offs, particularly between safety and availability when the malfunction occurs. Then, human/operator needs to shut-down the system. However, this situation could lead to traffic congestion and traffic collisions. Hackers also can take this opportunity to access and steal the data. In this case, perhaps there exists other potential trade-offs between each of dependability attributes? Perhaps, there could be also potential trade-offs between dependability aspects vs ethical concerns?
EVPS may potentially bias in making decision. For example, apart from emergency vehicles, perhaps the algorithm may prioritize e.g., “non-emergency vehicles” during its operation. Ensuring the fairness and providing explainability of each decision generated by the system are essential. Relying on only checking “what goes into a model” might not be sufficient. How can we address this kind of unfairness, which is, by definition, difficult to spot.
One issue neither raised in the complaints nor discussed by legal is about data retrieval and deletion. GDPR allows individuals to inquiry and request the delete their data. Scholars argue that personal data is any data that could be ‘singled out.’ Even if anonymised, with sufficient data points, any vehicle could be mapped to the individual. What if that data has already been used for training of a model? Can – and should – we delete selected data from a deployed model? An option might be to retrain the whole system from scratch; is this feasible?
Through T-Framework, stakeholders involved can address the above mentioned considerations with mutual ground.
Please check this page to understand more on how to implement T-Framework for this case study.