Resource management in distributed sensor networks is a challenging problem. This can be attributed to the fundamental tradeoff between the value of information contained in a distributed set of measurements versus the energy costs of acquiring measurements, fusing them into the conditional probability density function (pdf) and transmitting the updated conditional pdf. Communications is commonly the highest contributor among these costs, typically by orders of magnitude. Failure to consider this tradeoff can significantly reduce the operational lifetime of a sensor network. While a variety of methods have been proposed that treat a subset of these issues, the approaches are indirect and usually consider at most a single time step. In the context of object tracking with a distributed sensor network, we propose an approximate dynamic programming approach that integrates the value of information and the cost of transmitting data over a rolling time horizon. We formulate this tradeoff as a dynamic program and use an approximation based on a linearization of the sensor model about a nominal trajectory to simultaneously find a tractable solution to the leader node selection problem and the sensor subset selection problem. Simulation results demonstrate that the resulting algorithm can provide similar estimation performance to that of the common most informative sensor selection method for a fraction of the communication cost.

The problem of minimum cost in-network fusion of measurements, collected from distributed sensors via multihop routing is considered. A designated fusion center performs an optimal statistical-inference test on the correlated measurements, drawn from a Markov random field. Conditioned on the delivery of a sufficient statistic for inference to the fusion center, the structure of optimal routing and fusion is shown to be a Steiner tree on a transformed graph. This Steiner-tree reduction preserves the approximation ratio, which implies that any Steinertree approximation can be employed for minimum cost fusion with the same approximation ratio. The proposed fusion scheme involves routing packets of two types viz., raw measurements sent for local processing, and aggregates obtained on combining these processed values. The performance of heuristics for minimum cost fusion are evaluated through theory and simulations, showing a significant saving in routing costs, when compared to routing all the raw measurements to the fusion center.

[A graphical models perspective] Distributed inference methods developed for graphical models comprise a principled approach for data fusion in sensor networks. The application of these methods, however, requires some care due to a number of issues that are particular to sensor networks. Chief of among these are the distributed nature of computation and deployment coupled with communications bandwidth and energy constraints typical of many sensor networks. Additionally, information sharing in a sensor network necessarily involves approximation. Traditional measures of distortion are not sufficient to characterize the quality of approximation as they do not address in an explicit manner the resulting impact on inference which is at the core of many data fusion problems. While both graphical models and a distributed sensor network have network structures associated with them, the mapping is not one to one. All of these issues complicate the mapping of a particular inference problem to a given sensor network structure. Indeed, there may be a variety of mappings with very different characteristics with regard to computational complexity and utilization of resources. Nevertheless, it is the case that many of the powerful distributed inference methods have a role in information fusion for sensor networks. In this article we present an overview of research conducted by the authors that has

Abstract — The problem of routing of sensor observations for optimal detection of a Markov random field (MRF) at a designated fusion center is analyzed. Assuming that the correlation structure of the MRF is defined by the nearestneighbor dependency graph, routing schemes which minimize the total energy consumption are analyzed. It is shown that the optimal routing scheme involves data fusion at intermediate nodes and requires transmissions of two types viz., the raw sensor data and the aggregates of log-likelihood ratio (LLR). The raw data is transmitted among the neighbors in the dependency graph and local contributions to the LLR are computed. These local contributions are then aggregated and delivered to the fusion center. A 2-approximation routing algorithm (DFMRF) is proposed and it has a transmission multidigraph consisting of the dependency graph and the directed minimum spanning tree, with the directions toward the fusion center.

In the classical approach, the problem of distributed statistical inference and the problem of minimum cost routing of the measurements to the fusion center are treated separately. Such schemes cannot exploit the “inherent” saving in routing costs arising from data reduction in a sufficient statistic for inference. Our approach is to conduct in-network processing of the likelihood function which is the minimal sufficient statistic and deliver it to the fusion center for inference. We employ the Markov random field (MRF) model for spatial correlation of sensor data. The structure of the likelihood function is well known for a MRF from the famous Hammersley-Clifford theorem. Exploiting this structure, we show that the minimum cost routing for computation and delivery of the likelihood function is a Steiner tree on a transformed graph. This Steiner-tree reduction preserves the approximation ratio, which implies that any Steinertree approximation can be employed for minimum cost fusion with the same approximation ratio. In this chapter, we present an overview of this approach to minimum cost fusion.

Recent advances in sensor technology coupled with embedded systems and wireless networking has made it possible to deploy sensors for numerous applications including target tracking, environmental science, defense information, and security. Sensor scheduling, a process to allocate sensing resources by optimizing a performance metric over a future time-horizon under constraints, is an effective method to improve performance for such problems. This work investigates myopic (one step ahead) and non-myopic (multiple steps ahead) sensor scheduling algorithms for target tracking applications. Two methods of predicting tracker performance are developed that can be used for tar-get tracking applications. The first is covariance-based, and it can be used with covariance-based scheduler costs. The second is unscented transform-based and it can be used with arbitrary scheduler costs. In application, both methods give a significant improvement in tracking performance over the tracking performance without sensor scheduling. The use of non-myopic sensor scheduling is often restricted due to an exponential dependency of computational and memory requirements on the length of prediction horizon.

A distributed camera network allows for many compelling applications such as large-scale tracking or event detection. In most practical systems, resources are constrained. Although one would like to probe every camera at every time instant and store every frame, this is simply not feasible. Constraints arise from network bandwidth restrictions, I/O and disk usage from writing images, and CPU usage needed to extract features from the images. Assume that, due to resource constraints, only a subset of sensors can be probed at any given time unit. This paper examines the problem of selecting the “best ” subset of sensors to probe under some user-specified objective- e.g., detecting as much motion as possible. With this objective, we would like to probe a camera when we expect motion, but would not like to waste resources on a non-active camera. The main idea behind our approach is the use of sensor semantics to guide the scheduling of resources. We learn a dynamic probabilistic model of motion correlations between cameras, and use the model to guide resource allocation for our sensor network. Although previous work has leveraged probabilistic models for sensor-scheduling, our work is distinct in its focus on real-time building-monitoring using a camera network. We validate our approach on a sensor network of a dozen cameras spread throughout a university building, recording measurements of unscripted human activity over a two week period. We automatically learnt a semantic model of typical behaviors, and show that one can significantly improve efficiency of resource allocation by exploiting this model.

managing or coordinating the use of sensing resources in a distributed environment, is a multi-objective optimization problem. In our earlier work, we proposed MASM (Market-Architecture for Sensor Management), a market-based approach to allocate sensor resources in real-time to various resource requestors. MASM models the multi-objective sensor management problem as a combinatorial-auction based market where the network resources sell goods to the resource requestors. To allow the resource requestors to participate in the market, MASM grants “budgets ” to these resource requestors based on their priority to the overall mission. However, for a given budget, self-interested resource requestors or buyers can learn from market-data and adapt their bidding behavior. This paper presents results of an initial experimental study, where the learning behavior of resource requestors is modeled and their effect on market performance is examined. I.

Given the explosion in number and types of sensor nodes, the next generation of sensor management systems must focus on identifying and acquiring valuable information from this potential flood of sensor data. Thus an emerging problem is deciding what to produce, where, for whom, and when. Identifying and making tradeoffs involved in information production is a difficult problem that market-based systems can “solve ” by allowing user values, or utilities, to drive the selection process. Essentially this transforms the traditional “data driven ” approach (in which multiple sensors and information sources are used, with a focus on how to process the collected data) to a user-centered approach in which one or more users treat the information collection and distribution system as a market and vie to acquire goods and services (e.g., information collection, processing resources and network bandwidth). We describe our market-based approach to sensor management, and compare our prototype system to an information-theoretic system in a multi-sensor, multi-user simulation with promising results. This research is motivated in part, by rapid technology advances in network technology and in sensing. These advances allow near universal instrumentation and sensing with worldwide distribution. However while advances in service-oriented architectures and web-based tools have created “the plumbing ” for data distribution and access,

SEMARTCam scheduler: semantics driven real-time data collection from indoor camera networks to maximize event detection