Performance Evaluation of Neural Networks Applied to Queueing Allocation Problem
In this paper we consider the dynamic allocation of customers to queues and study the performance of neural networks applied to the problem. The queueing system consists of N parallel distinct servers, each of which has its own queue with infinite capacity. A controller allocates each arriving customer to one of the servers at arrival epoch, who maximizes the probability of starting service for the customer in the earliest time. A neural network is incorporated into the controller, so that the neural controller can make an allocation decision adaptively to changing situations. We present a simple training method for the neural controller. We consider two types of neural networks (BP and LVQ3) and compare their performance in numerical examples.
KeywordsService Time Arrival Rate Allocation Decision Service Time Distribution Neural Controller
Unable to display preview. Download preview PDF.
- L. Kleinrock, Queueing Systems, Vol. II: Computer Applications, John Wiley & Sons, NY, 1976.Google Scholar
- K. R. Krishnan, “Markov Decision Algorithms for Dynamic Routing,” IEEE Communications Magazine, pp. 66-69, October 1990.Google Scholar
- Z. P. Lo, Y. Yu, and B. Bavarian, “Derivation of Learning Vector Quantization Algorithms,” Proc. IJCNN’92, vol. III (Baltimore, MD), pp. III-561–566, June 7–11, 1992.Google Scholar
- D. E. Rumelhart, J. L. McClelland, and the PDP Research Group, Parallel Distributed Processing, Volume 1: Foundations, Cambridge, MIT Press, 1988.Google Scholar
- A. C. Tsoi and R. A. Pearson, “Comparison of Three Classification Techniques, CART, C4.5 and Multi-Layer Perceptrons,” Neural Information Processing Systems 3, pp. 963–969, Morgan Kaufmann Publishers, CA, 1991.Google Scholar