Abstract. In this paper, we present a method for approximating the values of sensors in a wireless sensor network based on time series forecasting. More specifically, our approach relies on autoregressive models built at each sensor to predict local readings. Nodes transmit these local models to a sink node, which uses them to predict sensor values without directly communicating with sensors. When needed, nodes send information about outlier readings and model updates to the sink. We show that this approach can dramatically reduce the amount of communication required to monitor the readings of all sensors in a network, and demonstrate that our approach provides provablycorrect, user-controllable error bounds on the predicted values of each sensor.

We propose an energy-efficient framework, called SAF, for approximate querying and clustering of nodes in a sensor network. SAF uses simple time series forecasting models to predict sensor readings. The idea is to build these local models at each node, transmit them to the root of the network (the "sink"), and use them to approximately answer user queries. Our approach dramatically reduces communication relative to previous approaches for querying sensor networks by exploiting properties of these local models, since each sensor communicates with the sink only when its local model varies due to changes in the underlying data distribution. In our experimental results performed on a trace of real data, we observed on average about 150 message transmissions from each sensor over a week (including the learning phase) to correctly predict temperatures to within +/-0.5• C. SAF also provides a mechanism to detect data similarities between nodes and organize nodes into clusters at the sink at no additional communication cost. This is again achieved by exploiting properties of our local time series models, and by means of a novel definition of data similarity between nodes that is based not on raw data but on the prediction values. Our clustering algorithm is both very efficient and provably optimal in the number of clusters. Our clusters have several interesting features: first, they can capture similarity between far away nodes that are not geographically adjacent; second, cluster membership to variations in sensors' local models; third, nodes within a cluster are not required to track the membership of other nodes in the cluster. We present a number of simulation-based experimental results that demonstrate these properties of SAF.

No abstract

We study the problem of providing a sensor node with an accurate estimate of the current time, from a novel prospective which is complementary to the well-studied clock synchronization problem. More precisely, we analyze the case in which a sensor node is temporarily unable to run a clock synchronization protocol due to low-energy, or intermittent connectivity, or process failures, but still requires an accurate estimate of the time. We propose and analyze two efficient clock reading methods, one deterministic and the other probabilistic, which are designed to work in synergy with clock synchronization protocols to provide a better time estimate. Our deterministic method achieves a better accuracy by exploiting the sign of the global deviation of the hardware clock from the reference time, and can be applied to reduce the frequency of the periodic clock adjustments by a factor 2, while maintaining the same time uncertainty. The second method based on time series forecasting, is more flexible than the previous one since it is independent of the frequency at which clock synchronization occur.

Ensuring the consistency and the availability of replicated data in highly mobile ad hoc networks is a challenging task because of the lack of a backbone infrastructure. Previous work provides strong data guarantees by limiting the motion and the speed of the mobile nodes during the entire system lifetime, and by relying on assumptions that are not realistic for most mobile applications. We provide a small set of mobility constraints that are sufficient to ensure strong data guarantees and that can be applied when nodes move along unknown paths and speed, and are sparsely distributed.In the second part of the paper we analyze the problem of conserving energy while ensuring strong data guarantees, using quorum system techniques. We devise a condition necessary for a quorum system to guarantee data consistency and data availability under our mobility model. This condition shows the unsuitability of previous quorum systems and is the basis for a novel class of quorum systems suitable for highly mobile networks, called Mobile Dissemination Quorum (MDQ) systems. We also show a MDQ system that is provably optimal in terms of communication cost by proposing an efficient implementation of a read/write atomic shared memory.The suitability of our mobility model and MDQ systems is validated through simulations using the random waypoint model and the restricted random waypoint on a city section. Finally, we apply our results to assist routing and coordinate the low duty cycle of mobile nodes while maintaining network connectivity.

Secure digital time-stamps play a crucial role in many applications that rely on the correctness of time-sensitive information. Well-known time-stamping systems are based on linking schemes which provide a relative temporal order by linking requests together. Unfortunately, these schemes do not scale well to large volume of requests, and have coarse granularity and high latency. Therefore, they are unsuitable for very large and dynamic systems, or applications requiring fine-grained and short-lived timestamps, or timeliness, such as stock trading, eauctions, financial applications, aggregation of real-time sensitive information, and temporal access control. We propose a different scheme, based on real-time timestamps, which overcomes those drawbacks and leads to a performance enhancement. Our time-stamping system is intrusion-tolerant and it is based on a novel robust time service suitable for very large populations, on a robust threshold signature scheme, and on quorum system techniques. We prove its correctness and liveness, and compare its computational and communication complexity to the complexity of tree-based linking scheme. Finally, we show how the fine-granularity, improved scalability and efficiency of our scheme make it particularly suitable to those applications mentioned above, and also to mobile e-commerce.

No abstract

Quorum systems are well-known techniques designed to enhance the performance of distributed systems, such as to reduce the access cost per operation, to balance the load, and to improve the system scalability. All of these properties make quorum systems particularly attractive for largescale sensor applications involving coordinated tasks, such as rescue applications. In this paper we analyze quorum techniques in the specific context of sensor networks and energy conservation, and show why quorum systems designed for wired networks and their metrics fail to address the challenges introduced by sensor networks. We then redefine quorum metrics such as access cost, load balance, and capacity in a way that takes into account the limitations and the characteristics of sensor networks, and discuss some energy-efficient design strategies. In addition, we propose a family of energy-efficient quorum systems and a particular construction, called Regional Quorum system (RQ), which reduces the quorum access cost. Finally, we propose a data diffusion protocol built on top of the RQ system, which improves energy consumption by reducing message transmissions and collisions, and increases the available bandwidth. We apply our diffusion protocol to analyze the RQ system using our novel metrics.International Conference on Wireless Algorithms, Systems and Applications 0-7695-2981-X/07 $25.00