Date: 19.08.2019 (Monday)
Time: 3.20 p.m.
Venue: ESB 244
Scholar: Konnur Sneha Basavaraj (EE15S034)
Guide: Dr. Krishna Jagannathan
Co-Guide: Gaurav Raina
Road traffic congestion is a major problem in cities today. There is a need to devise intelligent traffic signal control policies to handle the growing traffic scenario. In this work, we propose intelligent signal control policies for road transportation networks that increase network throughput and reduce delay. The signal control policies need to be scalable for large road networks, typical of a city. Hence, we focus on low complexity, distributed, vehicle actuated traffic signal control policies.
We propose variants of the well-known Backpressure algorithm. Our first algorithm – Backpressure algorithm using functions of queues, modifies the backpressure algorithm by using the difference in specific functions of Queues rather than just queue lengths. We prove throughput optimality for this algorithm. We also show that using functions of queues, we can enhance the performance for typical network scenarios
To avoid starvation of lightly loaded links, we propose delay-based policies namely, the Delay Backpressure algorithm and the Queue Delay Backpressure algorithm. The Delay Backpressure policy takes into account the delay experienced by the head of the line vehicle for computing backpressure. The queue delay backpressure algorithm uses both queue-lengths and delays to compute backpressure, allowing a trade-off between throughput and maximum delay incurred in the network.