Sandip Bapat

ExScal: elements of an extreme scale wireless sensor network

Project ExScal (for extreme scale) fielded a 1000+ node wireless sensor network and a 200+ node peer-to-peer ad hoc network of 802.11 devices in a 13km by 300m remote area in Florida, USA during December 2004. In comparison with previous deployments, the ExScal application is relatively complex and its networks are the largest ones of either type fielded to date. In this paper, we overview the key requirements of ExScal, the corresponding design of the hardware/software platform and application, and some results of our experiments.

Kansei: a testbed for sensing at scale

The Kansei testbed at the Ohio State University is designed to facilitate research on networked sensing applications at scale. Kansei embodies a unique combination of characteristics as a result of its design focus on sensing and scaling: (i) Heterogeneous hardware infrastructure with dedicated node resources for local computation, storage, data exfiltration and back-channel communication, to support complex experimentation, (ii) Time accurate hybrid simulation engine for simulating substantially larger arrays using testbed hardware resources, (iii) High fidelity sensor data generation and real-time data and event injection, (iv) Software components and associated job control language to support complex multi-tier experiments utilizing real hardware resources and data generation and simulation engines. In this paper, we present the elements of Kansei testbed architecture, including its hardware and software platforms as well as its hybrid simulation and sensor data generation engines

Analyzing the yield of ExScal, a large-scale wireless sensor network experiment

Recent experiments have taken steps towards realizing the vision of extremely large wireless sensor networks, the largest of these being ExScal, in which we deployed about 1200 nodes over a 1.3 km by 300 m open area. Such experiments remain especially challenging because of: (a) prior observations of failure of sensor network protocols to scale, due to network faults and their spatial and temporal variability, (b) complexity of protocol interaction, (c) lack of sufficient data about faults and variability, even at smaller scales, and (d) current inadequacy of simulation and analytical tools to predict sensor network protocol behavior. In this paper, we present detailed data about faults, both anticipated and unanticipated, in ExScal. We also evaluate the impact of these faults on ExScal as well as the design principles that enabled it to satisfy its application requirements despite these faults. We describe the important lessons learnt from the ExScal experiment and suggest services and tools as a further aid to future large scale network deployments.

Stabilizing reconfiguration in wireless sensor networks

A commonly desired feature of large-scale, multi-hop, wireless sensor networks (WSNs) is the ability to reconfigure them after deployment. This reconfiguration could be as simple as changing a single parameter or as complex as replacing the entire application. Several protocols have been proposed to enable reconfiguration in WSNs, most of which use version numbers to distinguish new configurations from old ones. Due to constraints on memory and message sizes, version numbers are bounded and use wraparound arithmetic to handle rollover. While this simple scheme works well in the common case, we identify in this paper, a serious version management problem in existing protocols due to which a reconfiguration operation may never stabilize. We analyze potential causes of this problem and its effects on the quality and lifetime of the network. Through extensive simulations and experiments, we demonstrate the significant likelihood of this problem occurring in real deployments. Finally, we provide a solution to this problem using a novel approach which we call human-in-the-loop stabilization. Our stabilizing reconfiguration protocol uses local detectors and correctors that can detect version inconsistencies and prevent their propagation in a timely and efficient manner, while ultimately allowing the human operator to restore the network to the correct configuration. Our simulations and experiments also demonstrate the performance benefits of our solution over previous, non-stabilizing protocols

Chowkidar: A Health Monitor for Wireless Sensor Network Testbeds

Wireless sensor network (WSN) testbeds are useful because they provide a way to test applications in an environment that makes it easy to deploy experiments, configure them statically or dynamically, and gather performance information. Sensor data collected in the field can be replayed on nodes, and new ways to process the data can be tested easily. Testbeds are rapidly growing in size, with hundreds or thousands of devices, and testbed services are also becoming richer and more complex. Due to their size and complexity, faults can (and do) occur in these testbeds, affecting the outcomes of experiments. Awareness of testbed health status is important to both testbed administrators charged with maintaining functional services, and users who prefer to use healthy devices and like to know if there are any failures during their experiments. Based on our experience with Kansei, a large WSN testbed at Ohio State, we identify use cases that motivate the design of Chowkidar, a health monitoring facility. Key among these are: monitoring as a service that operates independently of users to provide up-to-date testbed status information; monitoring of heterogeneous testbed devices and networks; distinguishing between node and interface failures; and diagnosing common-mode failures such as power supply or Ethernet hub failure. We present in this paper, a centralized and a distributed Chowkidar protocol that reliably monitor the health of large, heterogenous WSN testbeds. We present experimentally measured Chowkidar performance as well as real experiences and lessons learnt from the integration of Chowkidar with Kansei, including feedback from both testbed users and administrators who have found Chowkidar to be a useful tool for improving the accuracy and efficiency of testbed experimentation and maintenance.

Project exscal

Project ExScal (for Extreme Scale) fielded a 1000+ node wireless sensor network and a 200+ node ad hoc network of 802.11 devices in a 1.3km by 300m remote area in Florida during December 2004. In several respects, these networks are likely the largest deployed networks of either type to date. We overview here the key requirements of the project, describe briefly how they were met and experimentally tested, and provide a pointer to our experimental results.

Chowkidar: Reliable and scalable health monitoring for wireless sensor network testbeds

Wireless sensor network (WSN) testbeds are useful because they provide a way to test applications in an environment that makes it easy to deploy experiments, configure them statically or dynamically, and gather performance information. However, WSNs are typically composed of low-cost devices and tend to be unreliable, with failures a common phenomenon. Accurate knowledge of network health status, including nodes and links of each type, is critical for correctly configuring applications on WSN testbeds and for interpreting the data collected from them. In this article we present a stabilizing protocol, Chowkidar, that provides accurate and efficient network health monitoring in WSNs. Our approach adapts the well-known problem of message-passing rooted spanning tree construction and its use in propagation of information with feedback (PIF) for the case of a WSN. The Chowkidar protocol is initiated upon demand; that is, it does not involve ongoing maintenance, and it terminates with accurate results, including detection of failure and restart during the monitoring process. Chowkidar is distinguished from others in two important ways. Given the resource constraints of WSNs, it is message-efficient in that it uses only a few messages per node. Also, it tolerates ongoing node and link failure and node restart, in contrast to requiring that faults stop during convergence. We have implemented the Chowkidar protocol as part of enabling a network health status service that is tightly integrated with a remotely accessible wireless sensor network testbed, Kansei, at The Ohio State University. We present experimental results from this testbed that validate the correctness and performance of Chowkidar. We also report on initial experiences and lessons learnt from the integration of Chowkidar with Kansei, including feedback from both testbed users and administrators who have found Chowkidar to be a useful tool for improving the accuracy and efficiency of testbed experimentation and maintenance, and the need for well-defined policies to address issues such as minimizing interference with concurrently running experiments. Finally, we discuss extensions that enhance the functionality and usability of Chowkidar.

Implementing an autonomic architecture for fault-tolerance in a wireless sensor network testbed for at-scale experimentation

The wireless sensor networking (WSN) community has increasingly grown to rely on experimentation with large-scale test-beds as a means of verifying protocols, middleware and applications. These testbeds need to be highly available in order to support this community, but are themselves complex, and complex to manage, being prone to faults in hardware, software specification and software implementation. In this paper we report on our experience in designing Kansei, a WSN testbed for experimentation at scale, to be autonomic - i.e. self-healing and self-managing. We implement autonomic management in Kansei through an architecture that consists of a hierarchy of self-contained components, extended with detectors for discovering faults and correctors for subsequent stabilization. We find that our invariant based architecture is well suited for large complex systems with unpredictable fault model and its fault monitoring framework can be extended to include user programs.

Stabilizing health monitoring for wireless sensor networks

Wireless sensor networks (WSNs) comprised of low-cost devices tend to be unreliable, with failures a common phenomenon. Being able to accurately observe the network health status -- of nodes of each type and links of each type -- is essential to properly configure applications on WSN fabrics and to interpret the information collected from them. In this paper we study accurate network health monitoring in WSNs. Specifically, we reconsider the well-known problem of message-passing rooted spanning tree construction and its use in PIF (propagation of information with feedback) for the case of a WSN. We present a stabilizing protocol, Chowkidar, that is initiated upon demand; that is, it does not involve ongoing maintenance, and it terminates with accurate results, including detection of failure and restart during the monitoring process. Our protocol is distinguished from others in two important ways. Given the resource constraints of WSNs, it is message-efficient in that it uses only a few messages per node. And it tolerates ongoing node and link failure and node restart, in contrast to requiring that faults stop during convergence. We have implemented the protocol as part of enabling a network health status service that is tightly integrated with a remotely accessible wireless sensor network testbed, Kansei, at The Ohio State University. We report on experimental results.

Message efficient termination detection in wireless sensor networks

Execution of wireless sensor network (WSN) applications typically consists of a number of successive phases such as network reprogramming, localization, power management, health monitoring, and parameter updates. Termination detection of a phase is therefore a critical operation for a network manager to safely execute a new phase on some or all of the network nodes. In this paper, we reformulate the well-known problem of termination detection for WSNs, and present an automated, low-cost solution to the problem. Our algorithm, Reporter, exploits the reactive nature of WSN protocols as well as the broadcast communication model of WSNs. It detects termination accurately and (message) efficiently, using reports from only a small fraction of nodes in the network. It exploits existing network traffic to construct a routing tree to collect these reports at a base station, and thus reduces the control overhead of structure formation. Moreover, it has low computation and memory overhead. We have developed a TinyOS implementation of Reporter, which is easily composed with a number of existing WSN protocols. We provide detailed experimental performance results obtained on an indoor mote testbed which show that Reporter selects as few as 5 % of the total number of nodes in the network for collecting termination reports while preserving accuracy.