Abstract
Wireless sensor networks (WSNs) are often structured as a data collection tree to collect sensory data into the sink node. Sensor nodes in the middle of the tree will help relay data for those nodes further away from the sink. This sort of multi-hop communication is very essential for WSNs. However, it is challenging to perform multi-hop communication on extremely resource-constrained sensor platforms, especially under high bandwidth. For one thing, the limited hardware resources may limit such platforms to do without the Received Signal Strength Indication facility, which makes communication scheduling very important. For another thing, the extremely small built-in memory allows only very simple and carefully designed communication protocols. This thesis addresses these challenges and introduces various techniques to overcome them, including a Token-based TDMA scheduling scheme integrated with multi-channel communication and balance tree construction to minimize the schedule frame for high data rate requirement. We take the ECO sensor platform as an example and show the feasibility of the proposed techniques through real implementation. When compared with the EcoDAQ protocol, which is a fast one-hop data acquisition protocol designed specially for ECO, our proposed techniques gain a 71% improvement in data throughput.