On-demand wakeup in wireless sensor networks

Suresh Mathew, Purdue University

Abstract

In today’s wireless sensor networks (WSN), power consumption continues to be a major problem despite advances in power management techniques. In general, the most power hungry component in a wireless sensor node is the radio subsystem. Apart from outgoing data transmissions, the radio is also responsible for monitoring the wireless channel for incoming data packets intended for it. Most existing radio power management techniques are based on the general principle of duty-cycling with scheduled wakeup. In a scheduled wakeup scheme, the receiver node wakes up at regular intervals to check for any incoming transmissions. If there are no data transmissions, it goes back to sleep. For WSN applications characterized by a small number of sporadic transmissions per day (e.g., detecting rare events), analysis of the energy profile of scheduled wakeup shows that 85-90% of a node’s energy is spent on monitoring the wireless channel. This implies that even though there are only very few transmissions, a substantial portion of the node’s energy is spent on merely looking for wireless activity. In addition, the cost of time synchronization, which is required for scheduled wakeup schemes to work effectively, also contributes to the energy overhead of the node. This research focuses on reducing the energy overhead due to channel monitoring. The approach, which we term on-demand wakeup, wakes up the main radio only when there is data to be received; else it is asleep. This design employs a novel hardware component called the on-demand wakeup trigger (OWT) component at the receiver node. The OWT component consumes almost zero power and is always on. When the transmitter is ready to transmit data, it first sends out an RF beacon that is processed by the OWT component to generate a hardware interrupt. This interrupt serves as an RF trigger to wake up the receiver’s main radio. Since the transmitter and receiver are now active, data transmission ensues. Hence the energy consumed by the main radio is eliminated in on-demand wakeup. A Markov chain based power model was developed to estimate the improvement in the node lifetime obtained by on-demand wakeup. The hardware and software for on-demand wakeup have been designed and integrated with a commercially available wireless sensor node to illustrate and evaluate the concept. For applications characterized by a small number of sporadic transmissions per day and fast response time requirement, experimental results showed a 4-6X increase in lifetime for on-demand wakeup when compared to scheduled wakeup. Also, the additional energy overhead due to the OWT component in the on-demand wakeup scheme is negligible when compared to the channel monitoring energy overhead in scheduled wakeup schemes.

Degree

M.S.E.C.E.

Advisors

Raghunathan, Purdue University.

Subject Area

Computer Engineering|Electrical engineering

COinS