Adaptive admission control and packet scheduling schemes for QoS provisioning in multihop WiMAX networks

Abstract

Selection of Call Admission Control (CAC) and packet scheduling are crucial for multihop WiMAX networks to satisfy the Quality of Service (QoS) for end users. In this paper, we propose an adaptive CAC method and two different scheduling schemes for multihop WiMAX networks. The proposed CAC in the multihop Base Station (BS) reserves some bandwidth (BW) for the mobile users and changes the BW reservation adaptively based on most recent requests from the handover users. When there are few or no handover users exist in a network, the remaining reserved BW is allocated to low priority Best Effort (BE) users for effective BW utilization. While admitting the New Calls (NCs) or Handover Calls (HCs), the BS verifies both BW and multihop delay requirements to satisfy the QoS of the call. Next, we propose two downlink scheduling algorithms (P+E) and (P+TB) for the BS in multihop networks. The (P+E) scheduler combines the Priority and Earliest Due Date (EDD) scheduling methods, while the (P+TB) scheduler combines the Priority and Token Bucket (TB) scheduling methods. When the network is lightly or moderately loaded, the (P+E) scheduler performs well for both single and multihop users but the performance of real time services are highly affected under high load conditions. On the other hand, the (P+TB) scheduler has very good QoS performance for real time services under high load conditions and also has closer QoS performance to (P+E) scheduler for multihop users under low and moderate load conditions. The simulation results for CAC show that the proposed CAC have lower Call Drop Probability (CDP) for HCs than existing fixed BW reservation policy. The CAC admits more NCs when fewer number of HCs arriving at the BS. The simulation results for scheduling show that the (P+E) scheduler outperforms the (P+TB) scheduler when the system is moderately loaded and the (P+TB) scheduler outperforms the (P+E) scheduler when the system is fully loaded.