Criteria for the deployment of a heterogeneous linear WSN: Operability vs energy efficiency

Abstract

Linear Wireless Sensor Networks (LWSN) are typically used for monitoring infrastructure systems extending over several kilometers, where propagation conditions can be inhomogeneous in different sections of the network. Obstacles and reflecting objects contribute to bit errors in data packets, which take extra time and energy to recover. However, the impact of the inhomogeneities across LWSN sections on the time and energy performance of the entire network remains underreported. In addition, conventional methods of building LWSN by distributing relay nodes among inhomogeneous sections have failed to balance operability and energy efficiency criteria. This article examines the models of time and energy loss in LWSN sections showing different signal fading, represented by the Rician K-factor in the expressions for the bit error probability. On the basis of the dependences of time losses on inter-node distances, algorithms for deploying a heterogeneous LWSN are proposed according to the criterion of minimizing end-to-end network delay without restrictions on hardware and energy resources or when distributing a limited number of nodes. As an alternative to the efficiency criterion, we propose algorithms for balancing power consumption in a heterogeneous LWSN at an acceptable level of power loss without limiting the number of nodes or at the lowest level of power loss with a limit on the number of nodes. Experiments were carried out on CC2500 TI node transceivers in five inhomogeneous LWSN sections, where K-factor ranges from 5 to 100, in order to demonstrate the efficiency of the proposed algorithms. Results showed that the end-to-end delay and energy loss can be reduced by 1.36 and 2.4 times, respectively, compared to those in networks with uniformly distributed nodes. To further assess the operability and energy efficiency of the algorithms, a scheme is proposed based on the ratio of normalized concessions in the Pareto region. Additionally, the optimal number of nodes is determined when selecting a rational algorithm for deploying a heterogeneous LWSN.