Batch Forwarding in Wireless Sensor Networks
Batch processing is a well-known concept in computer science and widely used to improve performance in many applications. Similarly, to improve the delivery ratio and end2end delay in wireless sensor networks (WSNs), we propose to use batch forwarding instead of streaming. Our main goal in this paper is to evaluate and compare the performance of batch forwarding and that of streaming. Accordingly, we first develop analytical models under simplified assumptions. We then conduct extensive actual experiments using TinyOS and IRIS motes from Crossbow. Both analytical and actual experimental results show that batch forwarding significantly improves the efficiency and delivery radio, particularly in heavily loaded network environments. Despite some delay due to batch formation, batch forwarding also improves the overall end2end delay performance due to its efficient use of resources. As a result, batch forwarding is a viable data gathering solution and must be used in WSN applications that generate significant amount of sensing data and require high delivery ratio and less delay.