Because of acquired understanding of the sensor network environment and continual technological advancement, the software running on sensor nodes needs to be changed; that is, the sensor network needs to be reprogrammed. This paper discusses the issues involved: 100 percent delivery, relatively high bandwidth requirements, message collisions and congestion, concurrent senders, and the importance of energy efficiency for low-powered sensor nodes. It presents a code dissemination protocol, multihop network reprogramming protocol (MNP), to provide “a reliable and energy-efficient service to propagate new program code to all sensor nodes in the network, over wireless radio.”

Contributions include a sensor selection algorithm, “pipelining to enable fast data propagation,” and three innovative sleep strategies for energy conservation: contention sleep, noreq sleep, and init sleep. Another contribution is implementation and performance evaluations using the TinyOS platform and “TOSSIM, a discrete event simulator for TinyOS wireless sensor networks.” This includes the identification of optimal values of the three sleep parameters in different network densities, network sizes, and base station locations, and the use of these optimal values to observe the performance of MNP. This report is on experiments done on a preliminary version of MNP, not the latest version. This analysis goes on to identify and discuss more MNP-related issues.

Kulkarni and Wang present a perspective on the related work of others, enabling readers to see how this work fits in with other sensor network reprogramming research. The authors also determine that investigation is needed on the use of MNP in the dissemination of any data.