Prajakta Desai's Homepage

Prajakta Desai, PhD Candidate
Department of Electronic Engineering
La Trobe University
Contact E-mail:

Prajakta received the in information technology from the University of Pune, Maharashtra, India in 2003. Currently, she is a PhD candidate at La Trobe University in the area of Intelligent Transportation Systems. She has worked for Persistent Systems, India and Sybase Software, India. She has over six years of R&D experience.Her research interests include Artificial Intelligence and Data Structures.

Research Title: Cooperative Vehicles for Traffic Congestion Management: a Multiagent Approach

Research Abstract:
This research aims to alleviate the problem of road traffic congestion by enabling efficient distribution of vehicles in a road network. The approach called Congestion Avoidance and Route Allocation using Virtual Agent Negotiation (CARAVAN) is based on multiagent systems, wherein the vehicles act as intelligent agents (autonomous software entities) and undertake cooperative route allocation via inter-agent communication and negotiation. The vehicle agents in the local vicinity communicate with each other and undertake cooperative route allocation before every designated decision point (junction) along their route. The vehicle agents use intervehicular communication to exchange their route preference information and undertake distributed processing to arrive at an initial allocation of routes. The allocation is improved via trading of routes which takes place in the form of “deals”, enacted by every agent virtually and in an autonomous, iterative and distributed manner, thereby reducing the communication requirements. The deals, being simple in nature, incur low computation overhead but at the same time generate efficient route allocations. CARAVAN is an “anytime algorithm”- which means that it can be interrupted at anytime for obtaining the solution within the given time-frame. CARAVAN is evaluated by integrating VanetMobiSim, used for vehicular mobility simulation, and JADE, used to simulate the behaviour of vehicle agents. The algorithm is extensively investigated for a variety of small and large synthetic and real-road networks of varying topologies. Overall, CARAVAN offers 13% to 43% reduction in travel time (depending on the simulation conditions) as compared to the Shortest Path Algorithm. The performance of the algorithm in terms of travel time reduction is evaluated for a wide range of algorithmic, environmental and agent-related parameters and compared with other non-cooperative algorithms thereby demonstrating the adaptive nature of the algorithm and the ability of its local coordination strategy to contribute towards achieving regulation of the overall traffic.


Simulation Tools and Results