Optimalisation of intersection signal phasing with Wi-Fi CSI

As traffic congestion increases world-wide, the optimisation of signal phasing of intersections is more and more desired. This research proposes a sensor setup consisting of Wi-Fi CSI sensors to enable cheap, non-intrusive adaptive traffic lights to adapt signal phasing to traffic conditions. Traffic conditions are monitored by the setup by quantifying queue length by vehicle counting and control delay by driving time determination with Wi-Fi CSI. The former is already implemented and the latter proved promising but requires more experiment data to verify. These monitoring techniques serve as inputs to pre-timed, queue-actuated and deep Q-learning traffic controllers which proved to be able to reduce the average waiting time by 15.5%. Wi-Fi CSI sensors in the proposed setup therefore provide as a complete solution for adaptive traffic lights.