Energy Efficient Networking Protocols for Wireless Sensor ... - CiteSeerX

16 downloads 119223 Views 60KB Size Report
The advantages and disadvantages of different protocols for different ... a MAC protocol for wireless sensor networks must have the energy efficiency as one of ...
Energy Efficient Networking Protocols for Wireless Sensor Networks Sihui (Sue) Zhou, Ren Ping Liu, and Y Jay Guo, Senior Member, IEEE

Abstract— Energy efficiency is widely regarded as one of the major challenges for wireless sensor networks. In the literature, the issue is normally addressed from the viewpoint of a specific protocol layer or functionality, such as medium access control (MAC) or routing. In this paper, energy conservation in wireless sensor networks is treated with a holistic approach and examined across all protocol layers and functionalities including MAC, topology management, routing and sensor protocols. The advantages and disadvantages of different protocols for different layers are reviewed and discussed. Some recommendations on the employment of protocols for small, medium and large scale wireless sensor networks are presented.

I. INTRODUCTION Wireless sensor networks pose interesting challenges for networking research. Foremost among these is the development of long-lived wireless sensor networks in spite of energy constraints of individual nodes. Sensor nodes are expected to be battery equipped, and deployed in a variety of terrains. In some of these deployments, it may be feasible to harness energy from ambient sources, such as solar power, whereas in others such as climate monitoring in the canopies, sensor nodes may not be able to renew their energy resources. From the networking protocol design perspective, energy conservation in wireless sensor networks can reside in the Medium Access Control (MAC) layer, the topology control layer, and the network layer. At the MAC layer, energy is conserved by reducing idle listening, protocol overheads, and collisions. At topology control, power consumption is reduced by adaptive duty cycling, where nodes turn off their radios when they are not required for multihop communications. At the network layer, energy saving is achieved by energy aware routing, data aggregation, and meta-data negotiation. Many energy efficient networking protocols have been proposed for wireless sensor networks in recent years. Most of these protocols are in research areas such as MAC [314], topology control [15-20], and routing [21-31]. Despite some differences in designs, all proposals have a common goal: that is to provide better energy conservations for wireless sensor networks. CSIRO ICT Centre, Australia {sue.zhou, ren.liu, jay.guo}@csiro.au

In this paper, we review advances in energy efficient networking protocols for wireless sensor networks. Additionally, a comparison in term of energy and network performance is also presented. Specifically, Section II reviews efficient MAC protocols along with primary requirements for wireless sensor networks, and Section III overviews advances in topology management schemes. Section IV presents an overview of routing protocols, and finally conclusions are given in Section V. II. ENERGY EFFICIENT MEDIUM ACCESS CONTROL PROTOCOLS

The traditional requirements on MAC protocols for wired or wireless LANs are high throughput, low delay latency, fairness, and low overhead. Although energy efficiency has been considered in the design of wireless LAN protocols [1][2], it is an optional function and has not been a primary design goal. In wireless sensor networks, all sensor nodes are battery powered, and there is limited or no user interventions for recharging/replacing batteries. Therefore, a MAC protocol for wireless sensor networks must have the energy efficiency as one of the primary requirements. Furthermore, idle listening has been identified as one of the major power wastage [4][5][6] in wireless sensor networks. TABLE-I shows some popular transceiver power consumptions [4][5][6][12]. In most cases, the energy consumption of idle listening is similar to that of receiving, and 50% of that for transmitting. For some devices, such as RFM Transceiver TR3000, the power consumption for idle listening is as high as that for transmitting/receiving. Therefore, managing idle listening is regarded as one of the main ways to improve the power efficiency of a MAC protocol [4][5][6]. In recent years, there have been many MAC protocols proposed for wireless sensor networks [3-24]. Most of these proposals tried to control medium access as well as minimizing energy consumptions. They can generally be classified into four groups: duty cycle controlling, random access, fixed access, and hybrid. TABLE-I POWER CONSUMPTION OF SOME POPULAR TRANSCEIVERS Transceiver Transmit Receive Idle Sleep listening RFM Transceiver 30mW 12mW 12mW 60µW (TR1001) RFM Transceiver (TR1000) MC13191

14.88mW

12.5mW

12.5mW

16µW

81mW

100mW

1.4mW

94µW

A. Duty Cycle Control Based Protocols Duty cycle control based MAC protocols are often called low duty cycle or sleep-wakeup protocols. All duty cycle control MAC protocols have an aim to reduce the idle listening period. Nodes sleep most of time in the idle period, and only wake up periodically for activities. When a node is in the sleep state, it will turn the radio complete off to conserve energy. A time-based duty cycle frame is used by all nodes in the network. Figure-1 shows a typical duty cycle frame structure. Wake-up

Sleep

Wake-up

Sleep

Wake-up

Sleep

Figure-1: The structure of time-based duty cycle frame

There have been a lot of duty cycle control based MAC protocol proposals. The most referenced protocols are SMAC [4], TMAC [6], and PMAC [5]. • S-MAC S-MAC was proposed by Ye, Heidemann, and Estrin [4]. The assumption of S-MAC is that a wireless sensor network has ad-hoc architecture and nodes spend a large amount time in the idle state. In S-MAC, there is no separate signaling channel for controlling packets. Therefore, there is a requirement for local synchronization. Virtual clusters are formed to group nodes with the same sleep synchronization schedule. This allows S-MAC to be scaled well for a large network. Based on IEEE 802.11, S-MAC uses RTS/CTS/ACK for medium access control. Each node periodically sleeps and wakes up. When it wakes up, it listens to see if there are any packets for it. In the sleep state, it turns off its radio and sets a timer to wake itself up. Another way of saving power in SMAC is to use burst data instead of short messages. Data is transmitted in small segments and long bursts. S-MAC has been implemented on Mica Motes and the duty cycle can be configured from 1% to 99%. When the nodes in S-MAC have no sleeps, the protocol becomes IEEE 802.11. The amount of energy savings depends on the duty cycle. It is a trade-off with network performance, such as delay latency and throughput. S-MAC is able to gain significant energy savings compared with IEEE 802.11 without duty cycle control. • TMAC Proposed by Dam and Langendoen, T-MAC is an improvement to S-MAC [6]. It introduces an adaptive duty cycle: all messages are transmitted in variable length bursts and nodes sleep in between. The lengths of bursts are dynamically determined. Similar to S-MAC, there are active periods and sleep periods in a time-frame. An active period ends if there is no activity for a time period of Ta. Ta is the minimum listening time in the time-frame. All bursts will be transmitted at the beginning of the frame. The upper bound of transmitting length is the buffer capacity of the node. Nodes in T-MAC are also grouped into virtual clusters, and a synchronization scheme similar to S-MAC is used. Like

IEEE 802.11, T-MAC uses RTS/CTS/ACK for medium access control. It is shown in [6] that T-MAC is capable of saving as much as 98% power compared with a traditional protocol for light loads, and it saves 5 times more energy than S-MAC for variable loads. • PMAC The major difference between Pattern-MAC (PMAC) and S-MAC is that the former has an adaptive instead of fixed sleep-wakeup schedule [5]. The schedule is changed dynamically according to network traffic conditions. Another difference between PMAC and S-MAC is that a node can sleep for the whole time-frame if it believes that there are no networking activities. In PMAC, a sensor node acquires the networking activities information by a way of pattern. Based on this information, a node may decide to have a long sleep, which may last for several time-frames. The pattern is generated by sensing the environment when a node is awake and it is exchanged among neighbors. In PMAC the time-frame structure design has to accommodate the pattern exchange. Because PMAC is able to adaptively sleep/wakeup, it offers more energy savings under light loads, and higher throughput under heavy loads compared to S-MAC [5]. B. Random Access Protocols Random access protocols have been considered the most suitable for wireless sensor networks due to their selforganizing nature. There have been a lot of proposals in recent years [7-17]. Most of these protocols are CSMA or CSMA/CA based. The energy saving in these protocols is accomplished by collision minimizations and reductions of control packets. The most referenced random access protocols are B-MAC [7] and CSMA/ * [8]. • B-MAC Proposed by Polastre, Hill and Culler, B-MAC is a CSMA based random access protocol. Different from duty cycle controlling based MAC protocols, B-MAC uses an adaptive preamble sampling scheme to reduce energy consumption and idle listening. The main feature of B-MAC is that it provides a basic access control platform and other MAC protocols can be built on the top of it for customized and optimized energy savings as well as network performance. It is fully reconfigurable with a set of interfaces for the add-on MAC protocols. B-MAC is the default MAC protocol for UC Berkeley’s MICA-II on Tiny-OS. The performance benchmark has shown that B-MAC outperforms S-MAC with greater energy savings and network performance [7]. • CSMA/ p* Proposed by Tay, Jamieson, and Balakrishnan [8], CSMA/ p* is a non-persistent CSMA, with non-uniform probability distribution p that a node uses to randomly select contention slots. The authors established a mathematical model and derived the optimal distribution p*. Using p* to select a contention slot, a node can reach ρ

the optimal network performance. In CSMA/p*, there is no special energy saving mechanisms. Any energy saving mechanism, such as duty cycle and topology control can combine with the proposed protocol [8] to obtain better power savings than the originals. For example, SMAC/CSMA/p* has a better network performance and power savings than S-MAC. C. Fixed Access Protocols Fixed access protocols have been used in a lot of wireless networks. The most popular ones are GSM, CDMA, 3G, and Blue-tooth networks. There has been a belief that fixed access based MAC protocols are able to gain the highest energy savings in wireless sensor networks [13]. However, they suffer from the problems of scalability and flexibility. Nevertheless, there have been a lot of proposals in this category for wireless sensor networks [9][13]. ER-MAC is a distributed TDMA based MAC protocol. By using the concept of energy-criticality as a function of energy and traffic rate, it provides balanced energy consumption as well as network performance [9]. In this protocol, power savings is achieved by eliminating collisions and reducing the idle listening and the control packet overhead. Two essential parameters, named the residual energy level, and the traffic rate, must be obtained by a node through a period of learning. The protocol is accomplished by TDMA grouping and leader election. The group leader synchronizes the time-slot for the group members. When a group member is not in its time-slot, it goes to sleep to preserve power. The group leader is not permanent, and it will be changed according to the energy criticality. Because ER-MAC is a distributed TDMA with maintained local synchronization, we can see it provides good energy savings as well as good network performance. D. Hybrid Protocols There have been some hybrid proposals, which are aimed at combining the advantages of random access with that of fixed access. All these protocols divide the access channel into two. Control packets are sent in the random access channel, and data packets are transmitted in the fixed access channel. The control channel schedules the data access. The hybrid protocols can gain high energy savings than random access protocols and offer better scalability and flexibility than fixed access protocols. Main hybrid protocols include Z-MAC [10] and IEEE 802.15.4 [11]. • Z-MAC Proposed by Rhee, Warrier, Aia, and Min, Z-MAC is a combination of TDMA and CSMA [10]. The authors showed that Z-MAC achieves high channel utilization and low latency than pure TDMA and CSMA. Z-MAC uses CSMA as the baseline with a TDMA schedule as a “hint” to improve contention resolution.

Z-MAC uses the concept of owner slot. A node has a guaranteed access to its owner slot (TDMA style), and contention based access to other slots (CSMA style). In this way, collisions are reduced and better energy savings achieved. There are two basic components in Z-AMC. One is called neighbor discovery and slot assignment, and the other is called local framing and synchronization. In the neighbor discovery and slot assignment, a TDMA group will be formed and a node is given a slot. A time frame is decided by the local framing and synchronization. Z-MAC introduces a new flexible time-frame rule without the need of global synchronization. However, it needs to perform global clock synchronization once at the setup phase. Implemented using Mica-II on TinyOS [14], it is shown that Z-MAC is more energy efficient than B-MAC and S-MAC [10]. • IEEE 802.15.4 Finalized in Oct 2003, IEEE 802.15.4 defined a MAC protocol for low-rate Wireless Personal Area Networks (WPAN) [11]. One of targeted applications for IEEE 802.15.4 is the area of wireless sensor networks. IEEE 802.15.4 is a hybrid-based protocol. It has a super-frame structure shown in Figure-2.

Figure-2: Super-frame structure in IEEE 802.15.4

In the frame, a TDMA-based period is used for guaranteed access, and a contention-based period is used for non-guaranteed access. All nodes can switch off their radios and enter the sleep state in an inactive period. There is a coordinator that operating in the beaconed mode to maintain the synchronization of time-frames. However, IEEE 802.15.4 can also operate in ad-hoc based. In this case, there is only contention-based period in the timeframe. In the contention-based period, the traditional CSMA/CA is used for resolve the contention. There is no special design for energy conservation in IEEE 802.15.4 except a typical duty cycle controlling scheme III. TOPOLOGY MANAGEMENT In many sensor deployment contexts, it will be far easier to deploy a large number of nodes initially than to add additional nodes at a later date. If we use too few of the deployed nodes, the distance between neighboring nodes will be too great and the packet loss rate will increase or the energy required to transmit the data over the longer distances will be prohibitive. If we use all deployed nodes simultaneously, the system will consume unnecessary

energy at best and, at worst, the nodes may interfere with one another by congesting the channel. Topology Management exploits the resulting redundancy in order to extend system lifetime. A. Power Control One approach in reducing energy consumption has been to adaptively control the transmit power of the radio. The lazy scheduling proposed in Prabhakar et al. [15] transmits packets with the lowest possible transmit power for the longest possible time such that delay constraints are still met. Ramanathan and Rosales-Hain [16] proposed some distributed heuristics to adaptively adjust node transmit powers in response to topological changes caused by mobile nodes. This work assumes that a routing protocol runs at all times and provides basic neighbour information that is used to dynamically adjust transmit power. B. Adaptive Duty Cycle Current research on adaptive duty cycling can be broadly divided into three categories: general schemes of sleep and wake-up, low duty cycle combined with MAC protocols, and duty-cycle control through topology management. Protocols in the last category explore the benefits that a dense network provides for energy savings. The basic idea is to only power on a small number of nodes that are sufficient to maintain network connectivity. GAF [17] utilizes geographic location information, and divides the network into fixed square grids1. Within each grid, nodes are equivalent from the routing point of view, so only one node needs to be active at any given time. In SPAN [19], each node decides whether to sleep or join the backbone based on connectivity information supplied by a routing protocol. In ASCENT [20], which is an adaptive self-configuration topology mechanism, each node assesses its connectivity and adapts its participation in the multihop network topology based on the measured operating region. IV. ROUTING Routing in sensor networks differs from the Internet routing in several ways: First, sensor nodes are tightly power constrained. Second, radio channels are lossy and can be asymmetric. Third, sensor networks mostly deal with low data rate traffic. Fourth, sensor network dynamics and deployment scale preclude global schemes for routing. Fifth, it often includes geographical location information. These challenges have led to proliferation of solutions that optimize on various performance requirements. We start by a brief review of Ad Hoc routing protocols which are the predecessors of the sensor network routing protocols, followed by an investigation of sensornet specific 1

A recent work [18] proposed to use hexagonal mesh instead of the square grid to eliminate an unreachable corner.

route metrics. Depending on applications, sensor network routing protocols can be N-to-one data aggregation style, or One-to-N data dissemination style. For each of these research areas, we provide an overview of the representative solutions. A. Ad Hoc Routing Sensor network stems from Ad Hoc network research. Many sensor network routings are deeply rooted in Ad Hoc routing protocols. It is imperative to review the existing ad hoc routing protocols in order to solve the multihop routing problem in ad hoc networks [21]. DSDV is a loop-free hop-by-hop distance vector routing protocol requiring each node to periodically broadcast routing updates. DSR uses source routing rather than hopby-hop routing, with each packet to be routed carrying in its header a complete, ordered list of nodes through which the packet must pass. AODV is essentially a combination of both DSR and DSDV. It borrows the basic on-demand mechanism of Route Discovery and Route Maintenance from DSR, plus the use of hop-by-hop routing, sequence numbers, and periodic beacons from DSDV. B. Route Metrics The metric most commonly used by ad hoc routing protocols is minimum hop-count. However, many wireless links have intermediate loss ratios and minimizing the hopcount maximize the distance traveled by each hop, which is likely to minimize signal strength and maximize the loss ratio. De Couto et al [22] presented the expected transmission count metric (ETX), which finds highthroughput paths on multi-hop wireless networks. ETX minimizes the expected total number of packet transmissions (including retransmissions) required to successfully deliver a packet to the ultimate destination. The ETX metric incorporates the effects of link loss ratios, asymmetry in the loss ratios between the two directions of each link, and interference among the successive links of a path. Sensor nets typically involve many resource constrained nodes that are densely connected by low-power radios. In [23], MintRoute explores ETX connectivity analysis and neighborhood management. MintRoute achieves reliable routing on dense sensor networks with simple, low-power radios and limited storage. The routing problem for sensor networks differs substantially from that of traditional ad-hoc wireless networks because sensor nets typically involve many energy constrained nodes. Energy aware routing has been an interesting topic in Sensor network routing. Earlier works were to minimize the total consumed energy to reach the destination, which minimizes the energy consumed per unit flow or packet. If all the traffic is routed though the minimum energy path to the destination the nodes in that

path will be drain-out of batteries quickly while other nodes, which perhaps will be more power hungry if traffic is forwarded through them, will remain intact. Instead of trying to minimize the consumed energy, the performance objective of maximizing the lifetime of the system is considered as a linear programming problem in [24]. Energy Aware Routing [25] maintains a set of paths and chooses one based on a probabilistic fashion. The value of the probability is inversely proportional to the cost (or energy consumption) of the path. By having paths chosen at different times, the energy of any single path will not deplete quickly. This can achieve longer network lifetime as energy is dissipated more equally among all nodes. C. Data-Centric Routing Sensor networks are intended to collect and actuate on data about the physical world, hence their use is expected to be highly data-centric. Unlike traditional end-to-end networking techniques, routing and data management need to be performed jointly in sensor networks in order to maximize energy savings. Therefore, a significant networking component is to provide a flexible platform to build data management frameworks that use various application-specific data aggregation schemes. Directed Diffusion [26] is data-centric in that all communication is for named data. All nodes in a directed diffusion based network are application-aware. This enables diffusion to achieve energy savings by selecting empirically good paths and by caching and processing in-network. Diffusion adopts a declarative, publish/subscribe API that isolates data producers and consumers from the details of the underlying data dissemination algorithms. It uses a twophase algorithm where data consumers seek out data sources by flooding Interests, and then sources search to find the best path by setting up Gradients back to subscribers. TinyDB [27] is a query processing system with small footprint intended for highly resource-constrained mote sensor nodes. It provides a declarative interface for data collection and aggregation inspired by selection and aggregation facilities in database query languages. TinyDB’s aggregation service, Tag (Tiny Aggregation), operates on a routing tree structure, where the root is typically a base-station or other egress point to users, and leaves of the tree constitute all nodes in the network. However, a tree topology is not robust against node and communication failures, which are common in sensor networks. Synopsis Diffusion [28] is a general framework for achieving significantly more accurate and reliable answers by combining energy-efficient multi-path routing schemes with techniques that avoid double-counting. Synopsis diffusion avoids double-counting through the use of order- and duplicate-insensitive (ODI) synopses that

compactly summarize intermediate results during innetwork aggregation. D. Epidemic Routing Epidemic routing refers to network protocols that allow rapid dissemination of information from a source through purely local interactions. Such epidemic algorithms are used for task dissemination, route discovery, and for code propagation in Sensor network maintenance. In an epidemic algorithm, a message initiated from a source is re-broadcasted by neighboring nodes and extends out-ward, hop by hop, until the entire network is reached. Early work showed that in wireless networks, simple broadcast retransmission could easily lead to the broadcast storm problem, where competing broadcasts saturate the network. This observation led to work in probabilistic broadcasts [29]. The SPIN protocols presented in [30] efficiently disseminate information among sensors in an energyconstrained multihop wireless sensor network by using meta-data negotiations to eliminate redundant signaling. Borrowing techniques from the epidemic, Trickle [31] regulates itself using a local “polite gossip” to exchange code metadata. Each node periodically broadcasts metadata describing what code it has. However, if a mote hears gossip about identical metadata to its own, it stays quiet. When a mote hears old gossip, it triggers a code update, so the gossiper can be brought up to date. V. CONCLUSIONS In the MAC layer, we believe that duty cycle control based protocols are more suitable for low data rate non realtime applications, such as environmental monitoring. Similarly, random access protocols perform better when traffic loads are low in the networks. On the contrast, hybrid protocols are more customized designs. They can be tailed for both low data rate as well as high data rate applications. Topology management is useful in large scale dense sensor deployment. It helps optimise network formation in the sensor field. While power control can be useful, their advantages are less pronounced in sensor networks, because the power consumed by these low-power radios in idle state is of the same order of magnitude than the Tx or Rx state, so optimizations on transmit power are less important. Turning the radio off and putting the transceiver in sleep state via Adaptive Duty Cycle is essential to extend network lifetime. Among published routing protocols, source routing is the most widely adopted in Sensor Networks. The key advantage of source routing is that intermediate nodes do not need to maintain up-to-date routing information in order to route the packets they forward. This fact, coupled with the on-demand nature of the protocol, eliminates the need for the periodic route advertisement and neighbor detection packets present in other protocols. Tuning routing metrics

could further enhance network performance and/or achieved significant energy savings. With in-network processing, aggregation, querying, meta-data, dissemination techniques, energy efficient networking can be achieved. From the above review, it can be seen that most protocols were designed with special assumptions and for special applications with a few vertically integrated designs. A recent proposal [32] define a unified link level abstraction that provide much greater modularity to sensor net designs, and offer a unified platform for different MAC, topology management, and routing protocols.

[16]

[17]

[18]

[19]

ACKNOWLEDGMENT The authors would like to acknowledge the contribution of numerous colleagues in the Networking Technologies Laboratory and Wireless Technologies Laboratory in CSIRO ICT Centre for their constructive discussions

[20]

[21]

REFERENCE [1] [2] [3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12] [13]

[14] [15]

LAN MAN Standard Committee of the IEEE Computer Society, ANS/IEEE Std 802.11 1999 Edition, 1999 M. Enescu, HIPERLAN type 2 / MAC Layer, 12.3.1999 S. Singh, and C. S. Raghavendra, PAMAS- Power Aware MultiAccess protocol with Signalling for Ad Hoc Networks, ACM Computer Communications Review, 1999 W. Ye, J. Heidemann, D. Estrin, Medium Access Control With Coordinated Adaptive Sleeping for Wireless Sensor Networks, IEEE/ACM Trans. On Networking, Vol. 12. No. 3, June 2004 T. Zheng, S. Radhakrishnan, and V. Sarangan, PMAC: An adaptive energy-efficient MAC protocol for Wireless Sensor Networks, The 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS’05), April 2005, Denver, Colorado, USA T. Dam, K. Langendoen, An Adaptive Energy-Efficient MAC Protocol for Wireless Sensor Networks, ACM Sensys’03, Nov 2003, LA, California, USA J. Polastre, J. Hill, and D. Culler, Versatile Low Power Media Access for Wireless Sensor Networks, ACM SenSys’04, Nov 2004, Baltimore, Maryland, USA Y. C. Tay, K. Jamieson, and H. Balakrishnan, CollisionMininizing CSMA and Its Applications to Wireless Sensor Networks, IEEE Journal on Selected Areas in Communications, Vol. 22, No. 6, Aug 2004 R. Kannan, R. Kalidinai, and S. S. Iyengar, Energy and Rate Based MAC Protocol for Wireless Sensor Networks, ACM SIGMOD, Vol 32, No. 4, Dec 2003 I. Rhee, A. Warrier, M. Aia, and J. Min, Z-MAC: a Hybrid MAC for Wireless Sensor Networks, ACM SenSys’05, Nov 2005, San Diego, California, USA LAN/MAN Standards Committee of the IEEE Computer Society, IEEE standard for Information Technology – Telecommunications and information exchange between systems – Local and metropolitan area networks – Specific part 15.4, Oct 2003 http://www.freescale.com/files/rf_if/doc/data_sheet/MC13191DS. pdf I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, Wireless Sensor Networks : A Survey, Computer Networks, 38 (2002) pp 393-422 http://www.tinyos.net/ B. Prabhakar, E. Uysal-Biyikoglu, and A.E. Gamal, “EnergyEfficient Transmission over a Wireless Link Via Lazy Packet

[22]

[23]

[24] [25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

Scheduling,” Proc. 20th Ann. Joint Conf. IEEE Computer and Comm. Soc. (INFOCOM), pp. 386-394, Apr. 2001. R. Ramanathan and R. Rosales-Hain, “Topology Control of Multihop Wireless Networks Using Transmit Power Adjustment,” Proc. 19th Ann. Joint Conf. IEEE Computer and Comm. Soc. (INFOCOM), pp. 404-413, Mar. 2000. Y. Xu, J. Heidemann, and D. Estrin, “Geography-Informed Energy Conservation for Ad Hoc Routing,” Proc. Seventh Ann. ACM/IEEE Int’l Conf. Mobile Computing and Networking (MobiCom), pp. 70-84, July 2001. R.P. Liu, G. Rogers, and S. Zhou, “Honeycomb Architecture for Energy Conservation in Wireless Sensor Networks”, submitted for publication, 2006. B. Chen, K. Jamieson, H. Balakrishnan, and R. Morris, “Span: An Energy-Efficient Coordination Algorithm for Topology Maintenance in Ad Hoc Wireless Networks,” Proc. Seventh Ann. ACM/IEEE Int’l Conf. Mobile Computing and Networking (MobiCom), pp. 85-96, July 2001. A. Cerpa, D. Estrin, ASCENT: Adaptive Self-Configuring sEnsor Networks Topologies, IEEE Transactions on Mobile Computing, vol. 03, no. 3, pp. 272-285, Jul., 2004. J. Broch, D. A. Maltz, D. B. Johnson, Y. C. Hu, and J. Jetcheva. A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols. In Proc. of the ACM/IEEE MobiCom, October 1998. D.S.J.De Couto, D.Aguayo, J.Bicket, R.Morris "A HighThroughput Path Metric for Multi-Hop Wireless Routing", In Proceedings of MOBICOM 2003, San Diego. A. Woo, T. Tong, and D. Culler. Taming the Underlying Challenges of Reliable Multihop Routing in Sensor Networks. In Proceedings of the first international conference on Embedded networked sensor systems, (SenSys 2003) 2003. J. Chang and L. Tassiulas, “Energy conserving routing in wireless ad hoc networks”, IEEE Infocom, 2000, pp. 22-31. R. C. Shah and J. M. Rabaey. Energy aware routing for low energy ad hoc sensor networks. In Proc. IEEE Wireless Communications and Networking Conference (WCNC), Orlando, FL, March 2002. C. Intanagonwiwat, R. Govindan and D. Estrin "Directed Diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks", In Proceedings of the ACM/IEEE International Conference on Mobile Computing and Networking (Mobicom), pages 56–67, Boston, MA, USA, August 2000. S. Madden, M. Franklin, J. Hellerstein, and W. Hong. Tag: a tiny aggregation service for ad-hoc sensor networks. In OSDI, volume 1, Boston, MA, 2002. S. Nath, P. B. Gibbons, S. Seshan, and Z. Anderson, Synopsis diffusion for robust aggregation in sensor networks. In Proceedings of the Second ACM Conference on Embedded Networked Sensor Systems (SenSys), 2004. R. Chandra, V. Ramasubramanian, and K. P. Birman, Anonymous Gossip: Improving Multicast Reliability in Mobile Ad-Hoc Networks, International Conference on Distributed Computing Systems, 2001. W. Heinzelman, J. Kulik, and H. Balakrishnan, Adaptive Protocols for Information Dissemination in Wireless Sensor Networks, Proc. 5th ACM/IEEE Mobicom Conference (MobiCom '99), Seattle, WA, August, 1999. P. Levis, N. Patel, D. Culler, and S. Shenker. Trickle: A SelfRegulating Algorithm for Code Propagation and Maintenance in Wireless Sensor Networks. In First Symposium on Network Systems Design and Implementation (NSDI), Mar. 2004. J. Polastre, J. Hui, P. Levis, J. Zhao, D. Culler, S. Shenker, and I. Stoica. A unifying link abstraction for wireless sensor networks. In Proceedings of the Third ACM Conference on Embedded Networked Sensor Systems (SenSys 2005).