Precomputation for Multi-constrained QoS Routing in High-speed Networks

As one of the most challenging problems of the next-generation high-speed networks, quality-ofservice routing (QoSR) with multiple (k) constraints is an NP-complete problem. In this paper, we propose a multi-constrained energy functionbased precomputation algorithm, MEFPA. It cares each QoS weight to b degrees, and computes a number (B= 1 2 − − + k k b C ) of coefficient vectors uniformly distributed in the k-dimensional QoS metric space to construct B linear energy functions. Using each LEF, it then converts k QoS constraints to a single energy value. At last, it uses Dijkstra's algorithm to create B least energy trees, based on which the QoS routing table is created. We first analyze the performance of energy functions with k constraints, and give the method to determine the feasible and unfeasible areas for QoS requests in the k-dimensional QoS metric space. We then introduce our MEFPA for k-constrained routing with the computation complexity of O(B(m+n+nlogn)). Extensive simulations show that, with few coefficient vectors, this algorithm performs well in both absolute performance and competitive performance. In conclusion, for its high scalability, high performance and simplicity, MEFPA is a promising QoSR algorithm in the next-generation high-speed networks. Keywords—QoS routing, precomputation, scalability, linear energy function, performance evaluation