Network-Wide Broadcast Service in Wireless Sensor Networks

Introdcution

Network-wide broadcast is one of the most fundamental services in wireless sensor networks (WSNs) (Akyildiz, Su, Sankarasubramaniam & Cayirci, 2002). It facilitates sensor nodes to propagate messages across the whole network, serving a wide range of higher-level operations: During the networking configuration, control messages may be broadcast from the sink to all the sensor nodes; For data collection, the interest or query messages may be broadcast within the network; Upon observing an event, a sensor node may broadcast a message to coordinate with other sensor nodes for tracking the event and storing the sensed data; to name but a few. Hence, implementing an effective network-wide broadcast service is critical to the overall performance optimization of a WSN.

Flooding and gossiping (Akyildiz, Su, Sankarasubramaniam & Cayirci, 2002) are two commonly used broadcast approaches, though their basic forms are known inefficient. If we assume that all sensor nodes in the network are active during the broadcast process (referred to as all-node-active assumption), ideally every sensor node needs to receive and forward the broadcast message at most once. Significant efforts thus have been made toward enhancing the efficiency of the basic flooding or gossiping, while retaining their robustness in the presence of the error-prone transmissions (Guo, 2004; Kyasanur, Choudhury & Gupta, 2006; Stann, Heidemann, Shroff & Murtaza, 2006).

The all-node-active assumption is valid for wired networks and for many conventional multi-hop wireless networks. It however fails to capture the uniqueness of energy-constrained wireless sensor networks. In a WSN, the sensor nodes are often alternating between the dormant state and the active state (Gu, Hwang, He & Du, 2007; Liu, Zhao, Cheung & Guibas, 2004; Wang, Xing, Zhang, Lu, Pless & Gill, 2003; Yan, He & Stankovic, 2003); in the former, they go to sleep and thus consume little energy, while in the latter, they actively perform sensing tasks and communications, consuming significantly more energy (e.g., 56 mW for IEEE802.15.4 radio plus 6 to 15 mW for Atmel ATmega 128L micro-controller and possible sensing devices on a MicaZ sensor mote). Define duty-cycle as the ratio between the active period and the full active/dormant period. A low duty-cycle WSN clearly has a much longer lifetime for operation, but breaks the all-node-active assumption. In such a network, if the number of sensor nodes is very small, it may be possible to wake up all sensor nodes for message broadcast through global synchronization with the customized active/dormant schedules. For larger scale WSNs, however, synchronization itself remains an open problem. More importantly, the duty-cycle schedules are often optimized for the given application or deployment, and a broadcast service accommodating the schedules is thus expected for the cross-layer optimization of the overall system.

In this chapter, we revisit the network-wide broadcast problem in low duty-cycle WSNs. Their scale, together with their application/deployment-specific duty-cycles, renders the all-node-active assumption impractical. This in turn introduces a series of new challenges toward implementing the network-wide broadcast service. From a local viewpoint, since the neighbors of a node are not active simultaneously, a node would have to forward a message multiple times at different time instances; From a global viewpoint, since the topology is time-varying with no persistent connectivity, if not well-planned, the latency for a broadcast message to reach all sensor nodes can be significantly prolonged. The error-prone wireless links further aggravate these problems. The experiments later in this chapter have shown that, for ultra low duty-cycles (less than 0.4), a conventional broadcast strategy would simply fail to cover all the sensor nodes within an acceptable timeframe.