Deploying long-lived and cost-effective hybrid sensor networks
Introduction
This paper investigates the problem of network deployment in hybrid sensor/actuator networks. By hybrid sensor networks, we mean those networks consisting of both resource-rich and resource-impoverished sensor devices. The resource-rich devices, called micro-servers, are more expensive but have significantly greater bandwidth and energy capabilities compared to the low-cost, low-powered sensors. Such hybrid sensor networks have the potential to support the long-range and/or high-bandwidth communications required by data-intensive sensing applications using broadband networking standards such as 802.16 as well as the low-power, fine-grained sensing possible by smaller sensing devices. Examples of broadband sensor networking applications include time-elapsed imaging using video sensors for coastal monitoring, and speech analysis in home health care and cane-toad monitoring.
In the past couple of years, sensor networks research has addressed the development of sensor platforms [1], application domains [2], and communication paradigms [3], [4], [5], [6]. However, they neither exploited hybrid device capabilities such as out-of-band data communication channels nor explored anycast services for sensor networks.
Historically, large scale networks have evolved to encompass myriad types of network devices. The Internet today combines different devices such as routers, servers and hosts. Even the routers can be classified into different categories (e.g. into core routers and edge routers). For large scale sensor networks that may have thousands of nodes in the future, it is more realistic to have hierarchical models of network devices rather than flat ones. Such a sensor network involves a hybrid of resource-rich specialized nodes in conjunction with small sensor devices [7]. The resource-rich nodes provide service such as (i) long-range data communications, (ii) persistent data storage, or (iii) actuation. Examples of actuation would be re-charging or replacing small nodes whose energy has been depleted, imagers which can take photos or video when activated by sensors, sprinklers used for precision agriculture which can sprinkle water in badly parched areas etc. The resource-rich node can act as a data sink, and we call it a micro-server. Fig. 1 shows the hierarchical view of a hybrid sensor network. The lower tier consists of numerous inexpensive sensors, e.g. MICA2 (see Fig. 2) from CROSSBOW [8]; and the upper tier consists of many expensive but resource-rich micro-servers, e.g. STARGATE (see Fig. 2) from CROSSBOW.
The key challenge in building Ad-Hoc multi-hop sensor networks from small, low-powered sensor nodes are scalable, energy-efficient mechanisms for data dissemination. Previously proposed data routing protocols [3], [4], [5], [6] for sensor networks have not been designed to leverage the capabilities of hybrid devices. By exploiting resource-rich devices, the communication burden on smaller, energy, bandwidth, memory and computation-constrained sensor devices can be reduced. Consequently, these protocols may not be best suited for several applications of such hybrid sensor networks, which involve a multitude of mutually cooperative micro-servers.
Our thesis is that an anycast service, which routes sensor data to the nearest available micro-server, rather than to a single designated server, can provide significant improvements to the aforementioned data dissemination protocols for such applications and networks. The intuition is that you only care for the service, not which server provides it. The anycast service should be useful for several hybrid sensor applications.
Consider the case of mobile soldiers operating in a battlefield. The soldiers may be equipped with more powerful data transmitters (out of band higher-range radios) than sensors. It may be more effective to forward the information (e.g. enemy detection, land mine presence, convoy vehicles) to the nearest available soldier, who can forward it to the other soldiers, instead of sending it to all soldiers in the field. In a disaster recovery operation, several biochemical sensors may have been scattered, and multiple imagers (aerial or robotic) may be navigating the terrain. When biochemical sensors detect a toxic plume, this message just needs to go to the nearest imager (rather than a specific imager) which can act accordingly. In the example of Fig. 1, resource-impoverished MICA2 motes transfer data to one of the STARGATES, and the STARGATE can either handle the data or transfer it to interested parties using out-of-band transmission channel (e.g. WiFi) and other routing protocols, e.g. Ad-hoc On-demand Distance Vector (AODV) routing [9].
Our previous work [10] shows that the use of an anycast protocol provides significant gains in network performance, such as system lifetime and data latency, when added to existing data dissemination protocols, such as Directed Diffusion [4]. In this paper, we choose to analyze anycast as follows.
In this paper, we investigate some fundamental questions on hybrid sensor network deployment to support anycast communication.
- •
Given a number of micro-servers, how does the placement of them affect the lifetime of network?
- •
What is the benefit of introducing additional micro-servers into the network? Is it cost-effective to introduce these extra micro-servers?
To answer these two questions, we formulate an integer programming problem to study how the placement of micro-servers affect the lifetime of a hybrid sensor network using anycast communication. This optimization problem allows us to study the cost-benefit of using multiple micro-servers. Our cost model accounts for the variation in the cost and capability of network resources in a hybrid sensor network, such as bandwidth and energy consumption, as well as the spatio-temporal variation in network events. In particular, we find that the cost-effectiveness of micro-servers increases with the size of the network, thus making hybrid sensor networks a scalable solution. Although we study network deployment in the context of anycast communication, a similar methodology can also be applied to distributed storage and computation in hybrid sensor networks.
The rest of this paper is organized as follows. Section 2 provides an overview of the anycast communication model and other related work. Section 3 proposes an integer linear programming formulation of the network deployment problem and proves that the problem is NP-hard. Section 4 introduces a tabu-search algorithm to solve the problem efficiently. Section 5 presents an analysis to compare the lifetime differences and a cost analysis of different scenarios. Section 6 discusses our conclusions.
Section snippets
Related work
In this section, we provide an overview of our anycast mechanism and the other related work.
Cost model and optimization
In this section, we propose a model to investigate how the number of micro-servers and their placement affect the lifetime of a hybrid sensor network and prove that the problem is NP-hard. We define network lifetime as the cumulative active time of the network until the time when the depletion of the first sensor or micro-server happens.
A tabu-search algorithm
Since the combinatorial optimization problem introduced in Section 3 is NP-hard, it is very inefficient to solve the problem and achieve optimized solution. From our experience, we find that the maximum network size that the state-of-art commercial optimization package CPLEX [22] can handle efficiently is 20 nodes. Thus, the results produced by CPLEX are not very helpful for the deployment of a reasonable size network. We therefore develop an heuristic solution based on tabu-search [23].
Results and analysis
The mathematical model introduced in Section 3 enables us to study the effect of the number of micro-servers and their placements on the network lifetime of hybrid sensor networks utilizing anycast routing. Moreover, this model also allows us to study the financial cost-effectiveness and in particular to determine the most cost-effective combination of sensors and micro-servers in a hybrid sensor network. Furthermore, our scalability studies show that the cost effectiveness of hybrid sensor
Conclusions
In this paper, we considered the problem of network deployment for hybrid sensor networks, consisting of both resource-rich and resource-impoverished sensor devices.
We model the sensor network as a graph. We proposed an integer linear programming formulation to maximize network lifetime, proved that it is NP-hard, and introduced a tabu-search algorithm to answer some fundamental questions related to hybrid sensor network deployment—for a given number of micro-servers, what is the maximum
Acknowledgements
National ICT Australia is funded through the Australian Government’s Backing Australia’s Ability initiative, in part through the Australian Research Council. Nirupama Bulusu is supported by research and equipment grants from the National Science Foundation (Award CISE-RR-0423728), Tektronix, the PSU Foundation, and the Maseeh College of Engineering and Computer Science.
Wen Hu is a Ph.D. candidate at the School of Computer Science and Engineering at the University of New South Wales. He holds a M.S. degree from University of New South Wales and a B.Econ. degree from Zhongshan University, China. His research interests include sensor networks and wireless mobile networks. He is a student member of IEEE Communication Society. He is a recipient of National ICT Australia (NICTA) Endorsement and a member of Networks and Pervasive Computing (NPC) program at NICTA.
References (23)
- et al.
System architecture directions for networked sensors
- et al.
Energy-efficient computing for wildlife tracking: design tradeoffs and early experiences with zebranet
- et al.
Tag: a tiny aggregation service for ad-hoc sensor networks
ACM SIGOPS Operating Systems Review
(2002) - et al.
Directed diffusion for wireless sensor networking
IEEE/ACM Transactions on Networking
(2003) - et al.
A two-tier data dissemination model for large-scale wireless sensor networks
- et al.
Information-directed routing in ad hoc sensor networks
- et al.
Computation hierarchy for in-network processing
- Crossbow Technology, Inc., Available from:...
- N.W. Group, Ad hoc on-demand distance vector (aodv) routing, IETF RFC 3561. Available from:...
- W. Hu, N. Bulusu, S. Jha, A communication paradigm for hybrid sensor/actuator networks, in: Proceedings of the IEEE...
Cited by (36)
A confidential and DoS-resistant multi-hop code dissemination protocol for wireless sensor networks
2013, Computers and SecurityCitation Excerpt :These two types of attacks will try to deplete the energy reserves of the sensor nodes by making them repeatedly perform the authentication of invalid data (e.g., an invalid signature) or transmit code updates on accepting spurious requests. These attacks, which rely on physical node compromise, can be defended against with Trusted Platform Module (TPM) (Hu et al., August 2010) technology which can provide secure storage capabilities. Defending against physical node compromise has not been addressed in this paper.
Integrating data from heterogeneous wireless sensor networks based on LoraWan and ZigBee sensor nodes
2020, 2020 29th International Scientific Conference Electronics, ET 2020 - ProceedingsUtilization of Multi-Sink Architectures for Lifetime Maximization in Underwater Sensor Networks
2019, 2019 2nd IEEE Middle East and North Africa COMMunications Conference, MENACOMM 2019Multiple Sink Placement Strategy for Underwater Wireless Sensor Networks
2018, 2018 International Symposium on Networks, Computers and Communications, ISNCC 2018SCAN: A Sensor Deployment Algorithm in Hazardous Regions via Mobile Actuators
2016, IEEE Sensors JournalA leader based intrusion detection system for preventing intruder in heterogeneous wireless sensor network
2016, 2015 IEEE Bombay Section Symposium: Frontiers of Technology: Fuelling Prosperity of Planet and People, IBSS 2015
Wen Hu is a Ph.D. candidate at the School of Computer Science and Engineering at the University of New South Wales. He holds a M.S. degree from University of New South Wales and a B.Econ. degree from Zhongshan University, China. His research interests include sensor networks and wireless mobile networks. He is a student member of IEEE Communication Society. He is a recipient of National ICT Australia (NICTA) Endorsement and a member of Networks and Pervasive Computing (NPC) program at NICTA.
Chun Tung Chou is a Senior Lecturer at the School of Computer Science and Engineering, University of New South Wales, Sydney, Australia. He received his B.A. in Engineering Science from University of Oxford, UK and his Ph.D. in Control Engineering from University of Cambridge, UK. He has published over 60 articles in Computer Networking and Control Engineering. His current research interests are Sensor Networks, Wireless Networks, Content Distribution Networks and Traffic Engineering.
Sanjay K. Jha is an Associate Professor and the head of the Network Group at the School of Computer Science and Engineering at the University of New South Wales. He holds a Ph.D. degree from the University of Technology, Sydney, Australia. His research activities cover a wide range of topics in networking including Wireless Sensor Networks, Adhoc/Community wireless networks, Resilience/Quality of Service (QoS) in IP Networks, and Active/Programmable network. He has published extensively in high quality journals and conferences. He is also participating in the Network and Prevasive Computing (NPC) program of NICTA and the wireless networking project at the Smart Internet CRC. He is the principal author of the book Engineering Internet QoS and a co-editor of the book Wireless Sensor Networks: A Systems Perspective (TBP May 2005). He is a Member-at-Large, Technical Committee on Computer Communications (TCCC), IEEE Computer Society. He is also a member of editorial advisory board of Wiley/ACM International Journal of Network Management. He has severed on program committees of several conferences. He also served as the Technical Program Committee Chair of the IEEE Local Computer Networks (LCN2004) and ATNAC04 conferences, and co-chair of the Ements-1 worskhop.
Nirupama Bulusu is an Assistant Professor of Computer Science at Portland State University. She received her Ph.D. from University of California Los Angeles (UCLA) in 2002, her M.S. from the University of Southern California (USC) in 2000, and her B.Tech. from the Indian Institute of Technology, Madras (IIT Madras), all in Computer Science. Her research interests lie in hybrid, wireless sensor/actuator networks and in applying the technology to health care and monitoring of ecosystems. She is the primary editor of the forthcoming book, “Wireless Sensor Networks: A Systems Perspective” (Artech House, 2005), a member of the organizing, program committee or editorial board of numerous sensor networks workshops and journals, and a recipient of the Provost’s PSU Foundation Faculty Development Award and the Institute Merit Prize from IIT Madras.
- 1
Tel.: +61 2 9385 7203; fax: +61 2 9385 5995.
- 2
Tel.: +61 2 9385 7383; fax: +61 2 9385 5533.
- 3
Tel.: +1 503 725 2404; fax: +1 503 725 3211.
- 4
This paper is a comprehensive extension of our earlier work in [W. Hu, C.T. Chou, S. Jha, N. Bulusu, Deploying long-lived and cost-effective hybrid sensor networks, in: Proceedings of the First Workshop on Broadband Advanced Sensor Networks (BaseNets 2004), San Jose, CA, October 25, 2004].