Abstract
The perception layer of the Internet of Things (IoT), built upon Wireless Sensor Networks (WSNs), is critical for data acquisition but is highly vulnerable to cascading failures. The failure of a single sensor can redistribute traffic, overload neighbors, and trigger cascading failures that compromise the entire network. Reliability models that neglect cascading failures tend to overestimate system reliability, potentially misleading design and maintenance decisions. While several studies have modeled this phenomenon, they typically rely on abstract connectivity metrics without considering sensing coverage. This thesis presents an analytical modeling method for assessing the end-to-end communication reliability of WSNs considering the impact of overload-driven cascading failures as well as the coverage requirement. The load-dependent failure rates at the node level are modeled using a power-law relationship. Binary decision diagrams are used to derive exact analytical expressions for communication reliability, which is defined by the success of a two-phase communication process (infrastructure and application). The proposed method is demonstrated using two case studies under different coverage requirements, and node and link reliabilities. A comparative analysis is also performed for scenarios with and without cascading failures. The findings provide quantitative guidance for designing WSNs with appropriate coverage, redundancy, and capacity margins to improve robustness at the IoT perception layer under cascading failure conditions.