[Show abstract][Hide abstract] ABSTRACT: Sensor network monitoring and control are cur- rently addressed separately through specialized tools. However, the high degree of coupling of network state to the physical environment in which the network is deployed demands that users can monitor the network and respond to network state changes continuously. This paper presents the open-source Octopus vi- sualization and control tool. Octopus is a protocol-independent tool that provides live information about the network topology and sensor data in order to enable live debugging of deployed sensor networks. It enables operators to reconfigure the network behavior, such as switching between time-driven, event-driven, and query-driven modes or between awake and sleep modes of one, many, or all nodes through its graphical interface. Octopus also supports changing duty cycles of nodes, data reporting period, or sensing thresholds in event-driven networks. Recon- figuration of nodes is achieved through short request messages that support typical reconfiguration options without the overhead of epidemically sending new program images over the air. Our empirical tests showcase Octopus's capacity to debug application behavior and to characterize heterogeneous network performance under multiple settings, as a step towards establishing a rules database that relates data delivery to network-level parameters, and towards enabling autonomous network reconfiguration. I. INTRODUCTION Wireless sensor and actuator networks (WSN's) are a po- tentially revolutionary network paradigm, where tiny wireless modules equipped with sensors monitor a physical space and enable operators to react to specific events. These networks are highly coupled with their physical environment: events in the physical environment affect the network traffic patterns, and the network is a means for remotely initiating actions in the physical environment. The tight coupling of WSN's with their physical environments distinguishes them from conventional wireless networks, and demands a fresh approach to monitoring and control. Conventional network monitoring involves tracking of net- work health and diagnostic data, while conventional network control involves reconfiguration according to operator require- ments. In contrast, monitoring WSN's involves monitoring of both the network state and the sensor physical parameters. Controlling WSN's involves controlling both the network configuration and the behavior of the sensing application. Another feature of conventional monitoring and control tools is that they are completely separate from the running application. For instance, a WiFi network sniffer is com- pletely separated from any data transfer application over the WiFi network. In contrast, a sensor network is primarily a monitoring network. Its typical application objective is, in fact, to monitor and control the physical environment through the network. These features present new opportunities for integrating monitoring and control of the network and the application in a meaningful way for effective sensor network research and development.
Wireless Communications and Mobile Computing 01/2011; 11:1073-1091. · 0.86 Impact Factor