A Cross-Layer Optimization Approach for Efficient Data Gathering in Wireless Sensor Networks

Data collection techniques in wireless sensor networks (WSN) suffer from heavy congestion particularly at nodes closer to the sink node. In order to combat this problem, either complex MAC layer protocols have been proposed or non scalable data collection solutions have been designed. We propose a novel cross layer optimization approach that assumes a very simple MAC protocol and makes use of both routing and MAC layers information to reduce congestion, improve delivery ratio, and optimize energy usage. The proposed approach uses multiple disjoint collection trees, rooted from sink, with non overlapping duty cycles. At the MAC layer, we exploit the fact that nodes that are on different data collection trees need not to communicate with each other, hence the SMAC based wake up and sleep schedule for each tree is different. Existing multiple tree based data collection protocols have been designed primarily for fault tolerance or load balancing. For MAC layer part of our protocol, we have modified the SMAC code available in ns-2.28 to simulate our data collection scheme. Our scheme improves the data delivery ratio up to 40% for regular traffic, and reduces energy consumption by 30%.