Reliability Comparison Between Plain-Based and Cluster-Based Linear Wireless Sensor Networks

Abstract

Linear wireless sensor networks (LWSNs) are special WSNs for condition monitoring of infrastructures whose typologies are linear such as railways, gas, or oil pipelines. The reliability of LWSNs is crucial for such applications due to their safety-critical or mission-critical nature. To improve reliability, people propose cluster-based LWSNs in which, besides sensor nodes and sink nodes, some particular nodes called backbone nodes that can relay packets are adopted. The cluster-based LWSNs, which provide alternative paths for all sensor nodes, seem more reliable than plain-based LWSNs. However, the reliability analyses of plain-based and cluster-based LWSNs have yet to be studied together. In this article, we propose a hybrid model and algorithms based on decision trees to assess the connectivity reliability of both plain-based and cluster-based LWSNs. The reliability comparison of them is performed, and some interesting results are deduced, such as the cluster-based LWSN is only sometimes more reliable than a plain-based LWSN with a maximum jump factor (MJF) = 2. The result can be used as fundamental principles of LWSN design for high-cost performance.