An analytical model for bounded WSNs with unreliable cluster heads and links

In Wireless Sensor Networks (WSNs), performance and availability are important in providing Quality of Service (QoS). WSNs are prone to failures that may result from software and hardware malfunctions, battery drain, tampering and link failures. In addition, sensors are resource constrained in terms of inadequate processing capacity, limited storage memory and restricted power supply. Alternating sensor operations between sleep and active modes whilst saving energy, has also introduced more challenges to the performance and dependability of WSN systems. This paper therefore seeks to investigate performance and availability concerns resulting from link and node failures in addition to the buffer restrictions and challenges caused by alternating sensor operations between active and sleep states. First, a finite queue capacity analytical model integrating performance and availability in the presence of node and channel failures and repairs is developed. Using the model, we analyse performance and availability of WSNs in terms of network throughput, delays and dependability. Finally, the model is used to evaluate tradeoffs between performance metrics. The analytical solutions using Spectral Expansion and Kolmogorov Forward equations as well as the simulation results are in good agreement.