Optimizing the Throughput of an M/G/C/C Topological Network

An M/G/C/C state dependent queuing network approach models the behavior of entities (e.g., pedestrians and vehicles) flowing through a space. For a space consisting of various complex topological networks, the challenge is how to control such flow so that its throughput can be optimized. This paper discusses the throughput optimization of an M/G/C/C system using a network flow programming model. For this, we first decompose each available network and analyze its performance separately. The performance reports the effect of arrival rates to the throughput, blocking probability, expected number of pedestrians and expected travel time of each of the network. The best arrival rate of each network is then searched and fed to the network flow programming model to find the optimal arrival rates of source networks. The effect of the optimal arrival rates to the performance of the whole network is then analyzed and compared to that of using arbitrary arrival rates. Results show that with the right control of arrival rates, pedestrians can smoothly and efficiently be flowed through a network.