| Summary: | 碩士 === 國立東華大學 === 電機工程學系 === 101 === The purpose of this study is to propose a novel cross-region taxi dispatching system, that seeks to yield an optimal matching among customers and taxis. The system will shorten Customer Waiting Time (CWT), enhance Success matching Rate (SR), enable taxis to serve more customers and to reduce Empty Cruising Time (ECT).
How to effectively dispatch a taxi to a customer is a combinatorial optimization problem. The aforementioned statistical measurements CWT, SR and ECT are related to the Application Quality of Service (A-QoS) of the dispatching system and thus affect how much both customers and taxi drivers are satisfied with the system. The system proposed in the study thus aims to maximize the A-QoS.
Currently-implemented dispatching systems usually divide their service area into several matching regions, since taxis are used to stay in a certain region to wait for customers. Furthermore, a small matching region where an optimal assignment computation is carried out requires less load of computation and so expedites the matching process.
In the thesis, we adopt the （Multi Agent Assignment Algorithm）linear assignment method for optimal taxi-customer matching in each matching region. Also, for better performance, a customer can be assigned to not only a taxi in his matching region, but also one in the border of his neighboring matching regions, named cross-region taxi. We compare the performance of our system with that of three other representative taxi dispatching systems, cPAQ, STNU, and NtuCAB. In our system, three different priority arrangements for mixing the local taxis and cross-region taxis in the taxi-queue, namely, FIFO, Fixed Ratio, Local First, are studied. The results show that our system outperforms other systems in particular in the setting of high customer arrival rate.
|
|---|
