Impact of node failure for duty-cycle MAC protocol design in wireless sensor networks

The paper presents two failure detection mechanisms, implemented at the MAC layer for nodes used in wireless sensor networks, in order to cope with transient or permanent node failures. Specifically, the failure detection mechanisms are particularly suited for periodic workload applications and MAC protocols that employ four-way handshaking (RTS/CTS/nDATA/nACK) for collision avoidance. One failure detection mechanism is reminiscent of the Hello protocol that is sometimes implemented at the Network layer, and uses a fixed time threshold, while the other mechanism uses the RTS packets count as a metric for determining that a failure has occurred. The paper focuses on the impact of implementing a failure detection mechanism on the performance of the network, in addition to comparing the two failure mechanisms. Moreover, the effect of buffer lengths variations is investigated on the setting of the fault tolerance parameters and on the network performance.