Speaker1 : Yun Li
Title : Adaptive Delay and Energy Efficient Duty Cycle Control for IEEE 802.15.4 Home Area Networks
The current IEEE 802.15.4 medium access control (MAC) protocol lacks effective delay-sensitive data transmission and adaptive duty cycle control schemes to support applications in home area networks (HANs). In this paper, we propose a novel cluster based adaptive delay and energy efficient duty cycle control scheme (ADECon) for IEEE 802.15.4 beacon-enabled HANs. We first formulate a classical dynamic programming (DP) inventory control problem to find the optimal cluster transmission policy which provides delay-sensitive data transmission and achieves energy efficiency simultaneously. We further propose a suboptimal adaptive transmission policy based on the threshold structure of the optimal policy to reduce the computational complexity. Moreover, ADECon is proposed based on the suboptimal policy. The simulation results show that ADECon can effectively adjust the duty cycle according to cluster traffic to reduce the end-to-end delay and energy consumption.
Speaker2 : Xingyu Han
Title : Cross-layer delay-driven log-rule packet scheduling policy with multiple queues multiple channels
The research focuses on cross-layer (between network layer and physical layer) log-rule delay-driven packet scheduling policy in a multi-channel scenario with dynamic queues. The policy determines how to deploy channels to different queues, which is also known as scheduling, on a slot-by-slot basis with the information of instantaneous channel states, queue length and packet delays in each queue. Unlike classic Max-weight type and Weighted exponential-rule, weighted log-rule scheduling policies are proven to be radial sum-rate monotone as packet length of each queue grows, which indicates log-rule policy concentrates more on balancing each queue than reducing the queue with longest delay or length, optimising the overall packet loss probability and packet delay. Scheduling a dynamic number of queues is a new area to investigate in, which is more realistic than the ones with a fixed number of queues. Delay-driven policy is derived by queue-driven one. In a scenario with fixed number of queues, packet delay stays proportional to packet length after some initial settings based on Little’s Law, and weighted queue-driven policy is proven to be throughput optimal. However, within a scenario with dynamic number of queues, it has been proven that scheduling policy accounting for queue length is no longer throughput optimal, which requires a new kind of delay-driven scheduling policy to keep all queues within stability region as long as the average arrival rate for each queue is smaller than average service rate. The scheduling policy considering both packet delays and channel state information is low in complexity and it is new in literature.
Date: 19th June, 2013.
Time: 14.00-15.00 hrs
Venue: QMUL Maths:1.03