Mobile agent migration modeling and design for target tracking in wireless sensor networks
Introduction
In wireless sensor networks (WSNs), sensor nodes are normally scattered in an area of interest. When certain events occur, the alerted sensors will collect data and forward them to a processing center. The processing center processes the data and generates decision results, which can be accessed by users through some other networks, such as the Internet. Sensor networks can be used in a wide spectrum of applications, including battlefield surveillance, flood detection, home automation, etc. [1].
One of the unique features of sensor networks is the necessity of collaboration. Each sensor node normally has limited processing capability, constrained power usage, and limited sensing range. Therefore, collaboration among sensor nodes is important in order to compensate for each other’s capability as well as improve the degree of fault tolerance. The characteristics of sensor networks bring up some important issues for collaborative signal and information processing (CSIP) in sensor networks, which we summarize as follows:
- •
Dense deployment of sensor nodes [2], [3]. Since in the future, thousands of sensor nodes will be densely deployed in the field, it is possible for the sensor network to provide dense spatial sampling of phenomena of interest. Therefore, the challenge would be to combine the distributed data, first at each node and then with collaboration among the relevant devices in the network to produce meaningful global results. One of the biggest concerns in this process is the design of scalable CSIP algorithms to combat the large scale.
- •
Asynchronous property [3]. The distributed processing in a WSN typically is asynchronous, for example, in a sequential fusion center, the data from other sensor nodes may arrive out of order. This makes it necessary to design relevant signal and information processing and fusion algorithms in order to deal with the asynchronous executions.
- •
Energy efficiency [4], [5]. Sensor devices are battery operated. Once deployed, it is usually very difficult, if possible at all, to replace the battery. Therefore, how to save energy, how to prolong the lifetime of individual sensor nodes as well as that of the whole sensor network is a major challenge in sensor network research. The network must optimize the trade-off between fault tolerance and energy efficiency in signal processing, data fusion, querying, and routing tasks in order to meet the energy constraints and at the same time to achieve reliable performance. One measurement of energy efficiency is the lifetime of sensor network, which is the time from node deployment to the time when the first node is out of function because of energy depletion.
- •
Reliability issue [6]. Sensor nodes are often deployed in harsh, disastrous or inaccessible environment. Such environment makes sensor nodes prone to failures. In addition, the wireless transmission in sensor networks has high bit error rate (BER) and low bandwidth. Such dynamic network environments present another challenge to achieve fault-tolerance and reliability.
- •
Requirement of progressive accuracy [3]. As mentioned before, the sensor nodes are normally battery powered. The limited energy resource calls for the development of power-aware signal processing and communication methods to provide progressive accuracy, such that the collaboration process can be terminated anytime upon achieving the desired accuracy to conserve energy.
In this paper, our focus is on the development of distributed computing paradigms to support CSIP in sensor networks to combat the above challenges.
Section snippets
Computing paradigms for CSIP
In the context of sensor networks, computing paradigm refers to the information processing model deployed at the application layer of the protocol stack. The computing paradigms used in sensor networks can be mainly divided into two categories: the client/server paradigm and the mobile agent paradigm.
The client/server paradigm has been one of the most popular models adopted in distributed computing [7], [8]. In this paradigm, a server offers a set of services, resources, and methods needed for
The mobile agent itinerary
Although the mobile agent computing paradigm has shown great potential for collaborative computing in WSNs, the biggest hurdle in practical deployment of the mobile agent based computing is the dynamic determination of the mobile agent itinerary.
The mobile agent migration route, including the selection of nodes and the order of migration, determines the level of energy consumption, data fusion accuracy, agent migration time, and has a significant impact on the overall performance of the sensor
Dynamic mobile agent migration modeling
The static mobile agent planning (SMAP) is a centralized algorithm that determines the route of the mobile agent once for all before dispatching the mobile agent. It may not be well suited for the distributed and dynamic natures of the wireless sensor network. A more desirable mobile agent planning algorithm should have a distributed feature to dynamically determines the next sensor node the agent needs to migrate on the fly. Then a major issue arises naturally – how to dynamically determine
Mobile agent planning algorithms
In the design of MAP algorithms, we would like to choose an optimal itinerary that consumes the least amount of resources (energy and time) for fulfilling the collaborative processing task. In order to do so, some information is needed for effective mobile agent planning, including either the global information or the local information. The SMAP problem needs the global information in order to generate the optimal itinerary. The global information is a complete picture of the whole network.
Simulation and algorithm evaluation
We use target tracking as an application example to show the effect of using different itinerary planning algorithms in support of collaborative signal and information processing in sensor networks. Fig. 13 first illustrates the mobile agent itineraries generated using different algorithms from Section 5. Here, a vehicle moves through the sensor field from left to right. Circles represent the sensor nodes, and the solid circle represents the processing center (PE), where the mobile agent is
Discussion and conclusion
This paper focused on the discussion of distributed computing paradigms to support CSIP in sensor networks. The advantages of using mobile agent computing over the traditional client/server computing are discussed first. Then the mobile agent migration is modeled and three mobile agent itinerary planning algorithms, ISMAP, IDMAP, and P-IDMAP, are presented. We then designed several experiments to investigate the effect of different parameters on the performance of the algorithms. We showed,
Yingyue Xu is a Research Scientist at Intelligent Automation Inc., Rockville, MD, USA. He received the B.S. and M.S. degrees in electrical engineering from Tianjin University, Tianjin, China, in 1999 and 2001, respectively, and the Ph.D. degree in electrical engineering from University of Tennessee in 2005. His research interests are in wireless sensor networks, collaborative signal processing and target tracking. He is a member of IEEE and a member of Sigma Xi.
References (29)
- et al.
Wireless sensor networks: A survey
Computer Networks
(2002) - et al.
Intelligent mobile agents in large distributed autonomous cooperative systems
Journal of Systems and Software
(1999) - et al.
Distributed computing paradigms for collaborative signal and information processing in sensor networks
International Journal of Parallel and Distributed Computing
(2004) - D. Estrin, R. Govindan, J. Heidemann, S. Kumar, Next century challenges: Scalable coordination in sensor networks, in:...
- et al.
Collaborative signal and information processing in micro-sensor networks
IEEE Signal Processing Magazine
(2002) - J. Flinn, M. Satyanarayanan, Energy-aware adaptation for mobile applications, in: Proceedings of Symposium on Operating...
- et al.
Energy-aware wireless microsensor networks
IEEE Signal Processing Magazine
(2002) - et al.
RMST: Reliable data transport in sensor networks
- et al.
Understanding code mobility
IEEE Transactions on Software Engineering
(1998) - et al.
Mobile Computing
(1996)
Any Time, Anywhere Computing
Multi-resolution data integration using mobile agents in distributed sensor networks
IEEE Transactions on Systems Man and Cybernetics Part C: Applications and Reviews
Cited by (0)
Yingyue Xu is a Research Scientist at Intelligent Automation Inc., Rockville, MD, USA. He received the B.S. and M.S. degrees in electrical engineering from Tianjin University, Tianjin, China, in 1999 and 2001, respectively, and the Ph.D. degree in electrical engineering from University of Tennessee in 2005. His research interests are in wireless sensor networks, collaborative signal processing and target tracking. He is a member of IEEE and a member of Sigma Xi.
Hairong Qi is an associate professor at University of Tennessee, Knoxville, TN, USA. She received B.S. and M.S. degrees in computer science from Northern JiaoTong University, Beijing, China, in 1992 and 1995, respectively, and the Ph.D. degree in computer engineering from North Corolina State University, Raleigh, in 1999. Her current research interests are collaborative signal and information processing in sensor networks, biomedical imaging, and automatic target recognition. She is a senior member of IEEE and a member of Sigma Xi.
- 1
Fax: +1 865 974 5483.