Background: Traffic congestion is one of the most severe problems especially in metro
cities due to ever increasing number of vehicles on roads by 20% per year even with well-planned
road management system and sufficient infra.
Objectives: Most of the existing traffic signal controllers use fixed cycle type, giving a constant
green/red/yellow phase for each traffic signal cycle. These traditional controllers cannot adapt the
dynamics of traffic at real time which a traffic man can do.
Methods: Deploying traffic men at every traffic light junction is not feasible due to manpower
shortage and cost considerations. In this work a three input fuzzy controller is proposed which can
adapt the dynamics of real time traffic and reduce the congestion at the traffic light junction. Proposed
fuzzy controller has three inputs namely; queue length, arrival rate and peak hours and one
output parameter, time extension which is to be controlled by the use of the three input parameters.
Results: All four lanes have been allocated a fixed green signal time of 60 seconds at the start. Extension/
decrease of the green light is done dynamically with ±28 seconds. Compared to conventional
fixed cycle type, proposed approach gives a minimum improvement of 6% and a maximum
of 47% depending on various traffic conditions at the junction.
Conclusion: In terms of CO2 emission improvement of 20% and 42.12% and in terms of fuel consumption
improvement of 34.73% and 57.18% has been observed compared to UCONDES (Urban
CONgestion DEtection System) and OVMT (Original Vehicular Mobility Trace) respectively.