A distributed taxi advisory system

This paper presents a distributed taxi advisory dispatch system (TADS) that estimates regions where the demand for taxis exceed the supply of free taxis. Taxis are given advisories to move into these regions to ensure a better balance between number of waiting clients and free taxis overall. TADS performs the following tasks: (1) estimates system states (number of free taxis and waiting clients), (2) generates appropriate number of advisory tokens and, (3) distributes advisory tokens. We perform evaluation of our proposed system based on traces obtained from a large Singapore taxi company that operates more than 15,000 taxis. Our results show that TADS can reduce the number of clients with wait times longer than 60 minutes by over 30%.