Security in wireless sensor networks for health monitoring helmet with anomaly detection using power analysis and probabilistic model

Abstract

Litigation faced by the NFL has called for better prevention and understanding of concussions and other sports injuries. To achieve this, sports officials have turned to wireless sensor networks, or WSNs, in the form of helmet sensors that automatically report any harmful injuries to attendants on the sidelines. While this approach provides players with a greater assurance of safety and a faster response to their injuries, the security weaknesses of WSNs must be addressed. These systems, being not only recently developed but also highly resource-constrained, may be easily manipulated by those looking to gain information about players (a form of passive attack) or even attempting to remove them from the game through the sending of false reports (a form of active attack). To prevent attacks such as these, we propose a system that uses a modification of the AES-CCM protocol as well as a novel attack detection system that uses probabilistic methods to report any harmful behavior to the user. The system's power usage due to injury reports is compared to a probability model that is based on past research that recorded the likelihood of injury for the positions played in professional football. This system offers many advantages over conventional cryptography as it is a lightweight approach that costs few resources; individual helmet sensors need only send simple power reports to a central base station which uses on-the-grid power to conduct security analysis. Provided below is detail of the paper which describes the problem in greater detail, a section that details the system architecture, a section that explains the AES-CCM protocol, and an explanation of the probabilistic approach. This is followed by a security analysis that compares the approach to several other approaches found in the literature, and finally a conclusion.