Investigating MAC Power Consumption in Wireless Sensor Network JAVIER BONNY Supervised by Jun LUO, Department of Communication Systems Swiss Institute of Technology (EPFL) Lausanne, Switzerland javier.bonny jun.luo @epfl.ch February 7, 2005 Note: Project Report for the course Self-Organized Mobile Networks, Doctoral School of the I&C School, Winter semester 2004/2005, Prof. J.-P. Hubaux, EPFL. Abstract Because of the difficulty in recharging or replacing the battery of each node in a wireless sensor network, the energy efficiency of the system is a major issue. There are different ways to reduce energy consumption, one of them is to have a MAC protocol specifically designed for this goal. In this report, we first describe two proposed MAC protocols for sensor networks emphasizing their energy saving methods. Finally, we present our proposal, RTS Preambling which further improve energy saving. 1 Introduction A wireless sensor network is composed of numerous nodes distributed over an area to collect information. The sensor nodes communicate among themselves through the wireless channel to self-organize into a multi-hop network and forward the collected data towards one or more base stations. Each node has one or more sensors, embedded processors and low-power radios, and is normally battery operated. Typically, these nodes coordinate to perform a common task. Low power capacities of sensor nodes result in very limited coverage and communication range compared to other mobile devices. Hence, to successfully cover the target area, sensor networks are composed of large number of nodes. 1
2 PROPOSED MAC PROTOCOLS 2 A MAC layer protocol proposed for the sensor networks should comply with these distinguishing sensor network properties. The most important attribute being the energy efficiency. Another important attribute needed is the scalability to the change in network size, node density and topology. Other secondary attributes are fairness, latency, throughput and bandwidth utilization. There will be generally some tradeoff between those secondary attributes and the energy efficiency. In this report, we will first present the source of energy waste in a MAC protocol. Then in section 2, we present two proposed MAC protocol (S-MAC and B-MAC) and discuss their energy saving methods. In section 3, we present our proposal, RTS Preambling and shows how it can further improve energy saving. Finally in section 4, we present our conclusions and provide future direction to finalize our protocol. 1.1 Energy Waste in MAC protocol The major sources of energy waste in a MAC protocol are the following: Collision: When a transmitted packet is corrupted it has to be discarded, and the follow-on retransmissions increase energy consumption. Control Packet Overhead: Sending and receiving control packets consumes energy too, and less useful data packets can be transmitted. Idle Listening: listening to receive possible traffic that is not sent. Overhearing: meaning that a node picks up packets that are destined to other nodes. Collision and packet overhead are common issues and should be addressed by every MAC protocol. We will thus not treat them in this report. We will put our focus on idle listening and overhearing, which are issues really important in a wireless sensor network. 2 Proposed MAC protocols In this section, we present S-MAC and B-MAC. We will just focus our attention in their energy saving methods and not present all their others important design achievements. 2.1 Sensor Mac (S-MAC) Locally managed synchronizations and periodic sleep-listen schedules based on these synchronizations are the basic idea behind the Sensor-MAC (S-MAC) protocol [1]. Neighboring cells form virtual clusters to set up a common sleep schedule. If two neighboring nodes reside in two different virtual clusters, then they wake up at listen intervals of both clusters. A drawback of S-MAC algorithm is this possibility to obey to two different schedules which results in more energy consumption by idle listening and overhearing.
2 PROPOSED MAC PROTOCOLS 3 Collision avoidance is achieved by carrier sense and RTS/CTS packet exchanges as in 802.11 standard. However, this scheme is only used in unicast communication. Periodic sleep may result in high latency especially for multi-hop routing algorithms, since all immediate nodes have their own sleep schedules. The latency caused by periodic sleeping is called sleep delay in [1]. Adaptive listening technique is proposed to improve the sleep delay, and thus the overall latency. In that technique, the node who overhears its neighbors transmissions wakes up for a short time at the end of the transmission. Hence, if the node is the next-hop node, its neighbor could immediately pass data immediately. The end of the transmissions is known by the duration field of RTS/CTS packets. Advantages: The energy waste caused by idle listening is reduced by sleep schedules. Besides its implementation simplicity, global time synchronization overhead may be prevented with sleep schedule announcements. Drawbacks: Broadcast communication does not use RTS/CTS, which increases collision probability. Adaptive listening incurs overhearing/idle listening if the packet will not be destined to the listening node. Sleep and listen periods are predefined and constant which decreases the efficiency of the algorithm under variable traffic load. 2.2 B-MAC B-MAC is a carrier sense media access protocol for wireless sensor networks that provides a flexible interface to obtain ultra low power operation, effective collision avoidance, and high channel utilization [2]. To achieve low power operation, B-MAC employs an adaptive preamble sampling scheme to reduce duty cycle and minimize idle listening. B-MAC supports on-the-fly reconfiguration and provides bidirectional interfaces for system services to optimize performance, whether it be for throughput, latency, or power conservation. B-MAC duty cycles the radio through periodic channel sampling that are called Low Power Listening (LPL). Each time the node wakes up, it turns on the radio and checks for activity. If activity is detected, the node powers up and stays awake for the time required to receive the incoming packet. After reception, the node returns to sleep. If no packet is received (a false positive), a timeout forces the node back to sleep. Accurate channel assessment (CCA) is critical to achieving low power operation with this method. Noise floor estimation of B-MAC is used not only for finding a clear channel on transmission but also for determining if the channel is active during LPL. False positives in the CCA algorithm (such as those caused by thresholding) severely affect the duty cycle of LPL due to increased idle listening. To reliably receive data, the preamble length is matched to the interval that the channel is checked for activity. If the channel is checked every 100 ms, the preamble must be at least 100 ms long for a node to wake up, detect activity on the channel, receive the preamble, and then receive the message. Idle listening occurs when the node wakes up to sample the channel and there is no activity. The interval between LPL samples is maximized so that the time spent sampling the channel is minimized. Advantages: Idle Listening is reduced to a minimum. It has a better overall performance than S-MAC.
3 OUR PROPOSAL: RTS PREAMBLING 4 Drawbacks: Overhearing issue is not solved. A long preamble increases the power consumption of all nodes in the sender s transmission coverage because of it. The duty cycle and thus the preamble length are tunable, but the sender and receiver should be tuned together. This requires a loose synchronization that is not easily achieved in a wireless sensor network. B-MAC is included in TinyOS since version 1.1.3 and thus is becoming the standard MAC protocol for sensor network. 3 Our Proposal: RTS Preambling Our proposal is based on B-MAC, it will thus inherit all of its qualities that have made it a standard for the sensor network research community. In addition we added an overhearing avoidance scheme to further improve its energy saving performance. 3.1 Idea The basic idea of RTS Preambling, as its name indicates, is to send useful information (RTS) in the preamble, instead of a constant as in B-MAC, in order to reduce overhearing. Nodes that are not the receiver after hearing a full RTS will go to sleep until the end of this transmission. In Figure 1, an example of execution is presented. The Sender will send as a preamble for its Data packet repeated RTS. In between two consecutive RTS, it will switch its radio to receive mode and listen for the channel for a DIFS period to be able to receive a CTS from the Receiver. A node when waking up from its sleep period will listen to the channel for a period L. This period L has to be longer than a DIFS to be certain to listen to a emitted RTS. The basic rule of preamble sampling is applied: when a node hears something it will continue to listen until the channel becomes free. In other words if a node is listening when a RTS is emitted, it will listen until it has received the full RTS. After having received this full RTS, the node can determine if it is the intended receiver or not. If it is not the case, it knows that it doesn t have to continue to listen to the channel and can go back to sleep until the end of the transmission. This sleep period T is indeed a NAV and is specified in the RTS packet. If it is the intended receiver, it will answer with a CTS after waiting for a SIFS period. After sending the CTS, it waits for another SIFS and turns it radio to receive mode to receive the Data packet sent by the Sender. When all the Data has been received, it can send an acknowledgement after waiting for another SIFS. 3.2 Comments The performance of the radio will determine how to set the different IFS period. How to set the NAV is still a pending issue. Since the sender doesn t know how many RTS it will have to send before receiving a CTS, determining the value for the NAV is not a trivial question. The RTS Preambling is good only for unicast. When we want to do a broadcast or a multicast, we must use an alternative. One of the possibility that we are investigating
4 CONCLUSION AND FUTURE WORK 5 Figure 1: RTS Preambling example is to repeat the data in the preamble. 4 Conclusion and Future Work With the RTS Preambling scheme, we avoid overhearing. The preamble used (RTSs, CTS, DIFS and SIFS) is no longer than the B-MAC preamble. Thus we can expect to achieve a better energy saving performance than B-MAC. Nevertheless before to make this conclusion we have some more steps to take. The first one being to figure out how to set the NAV. Then we should completely specify the alternative scheme for broadcast/multicast. And finally we must finish the implementation of the RTS Preambling scheme and then do some simulations and comparisons with B-MAC. References [1] Wei Ye, John Heidemann and Deborah Estrin. An Energy-Efficient MAC Protocol for Wireless Sensor Networks, In Proceedings of the 21st International Annual Joint Conference of the IEEE Computer and Communications Societies (INFO- COM 2002), New York, NY, USA, June, 2002 [2] Joseph Polastre, Jason Hill, David Culler. Versatile Low Power Media Access for Wireless Sensor Networks, In Proceedings of the Second ACM Conference on Embedded Networked Sensor Systems (SenSys 04), November 3-5, 2004.