COBRA A Multi-path Adaptive Local Load Sensing Routing Protocol for Wireless Sensor Networks

摘要

We propose a family of location-based load balancing routing algorithms for wireless sensor networks (WSNs) called COBRA. In this approach, we combine local load balancing with shortest path routing to strike a balance be- tween hop-count efficiency and load distribution. COBRA''s advantages over traditional routing protocols are: (i) a sig- nificant reduction of hot-spot occurrences (i.e., nodes or ar- eas of high congestion level), thus enabling longer unin- terrupted operation of WSNs, (ii) a light-weight and scal- able implementation that requires only local congestion up- dates, ideal for implementation on computationally chal- lenged platforms, and (iii) combination of routing and con- gestion avoidance tasks in the protocol stack--a feature that leads to smoother TCP operation over wireless channels with fewer false positives on existing congestion and sub- sequent sending rate reduction by TCP. We show through simulation that COBRA''s local load sensing in combina- tion with a shortest path routing component is a promis- ing paradigm that improves global performance in terms of throughput and scalability, while matching other protocols in the traditional energy efficiency objective.