Dynamic Scheduling in High Speed Downlink Packet Access Networks: Heuristic Approach

In this paper, a heuristic approach to build a near-optimal dynamic code allocation policy for the High-Speed Downlink Packet Access (HSDPA) downlink scheduler is presented. The approach is developed based on information about the structure of the optimal policy for the two-user case and then generalized (using results from order theory) to support any number of users. To find the optimal code allocation policy, an MDP-based dynamic programming model for the HSDPA downlink scheduler was developed. The model then solved using value iteration for the two-user case. The computation complexity of this model grows exponentially with the buffer size of each user and even more rapidly with the number of users. On the other hand, the proposed heuristic approach has linear computation complexity and can be extended to any finite number of users with any finite buffer size. Simulation is used to compare the performance of this policy to the performance of the optimal policy. The obtained results show that the heuristic policy performs very close to the optimal one with huge reduction in the computation time.