Local Read-Write Operations in Sensor Networks

TL;DR

This paper explores a local read-write abstraction for sensor networks, demonstrating its atomicity, efficiency advantages over similar operations, and analyzing its performance through experiments under various conditions.

Contribution

It introduces and evaluates a new local read-write abstraction for sensor networks, highlighting its atomicity and efficiency benefits over existing lightweight operations.

Findings

Local read-write is an atomic operation for sensor network synchronization.

It outperforms similar lightweight operations in optimistic scenarios.

Experimental results show performance varies with timeout settings and neighborhood size.

Abstract

Designing protocols and formulating convenient programming units of abstraction for sensor networks is challenging due to communication errors and platform constraints. This paper investigates properties and implementation reliability for a \emph{local read-write} abstraction. Local read-write is inspired by the class of read-modify-write operations defined for shared-memory multiprocessor architectures. The class of read-modify-write operations is important in solving consensus and related synchronization problems for concurrency control. Local read-write is shown to be an atomic abstraction for synchronizing neighborhood states in sensor networks. The paper compares local read-write to similar lightweight operations in wireless sensor networks, such as read-all, write-all, and a transaction-based abstraction: for some optimistic scenarios, local read-write is a more efficient neighborhood operation. A partial implementation is described, which shows that three outcomes characterize operation response: success, failure, and cancel. A failure response indicates possible inconsistency for the operation result, which is the result of a timeout event at the operation's initiator. The paper presents experimental results on operation performance with different timeout values and situations of no contention, with some tests also on various neighborhood sizes.