Matthias Ringwald

Passive inspection of sensor networks

Deployment of sensor networks in real-world settings is a labor-intensive and cumbersome task: environmental influences often trigger problems that are difficult to track down due to limited visibility of the network state. In this paper we present a framework for passive inspection (i.e., no instrumentation of sensor nodes required) of deployed sensor networks and show how this framework can be used to inspect data gathering applications. The basic approach is to temporarily install a distributed network sniffer alongside the inspected sensor network, with overheard messages being analyzed by a data stream processor and network state being displayed in a graphical user interface. Our tool can be flexibly applied to different sensor network operating systems and protocol stacks, and can deal well with incomplete information.

Next-generation prototyping of sensor networks

Large-scale deployment of sensor networks is more and more becoming an issue to researchers and industry alike. The recently revised BTnode architecture provides two wireless radios and facilitates the interconnection of heterogeneous devices. Apart from offering interesting new opportunities in using multi-frontend devices in sensor-network research, this architecture is optimally suited for deployment-support networks as introduced in the following.

Deployment of Sensor Networks: Problems and Passive Inspection

Deployment of sensor networks is concerned with setting up an operational wireless sensor network in a real-world setting. Unfortunately, deployment is a labor-intensive and cumbersome task as environmental influences often degrade performance or trigger bugs in the sensor network that could not be observed during lab tests. In this paper, we, firstly, study existing sensor networks to identify and classify typical problems that have been encountered during deployment. Secondly, we investigate whether and how the existence of these problems can be detected by means of passive inspection, where messages exchanged in the sensor network are overheard and analyzed such that modification of the sensor network is not required. We, thirdly, show how passive inspection can be implemented in a practical tool.

Deployment Techniques for Sensor Networks

The prominent visions of wireless sensor networks that appeared about a decade ago have spurred enormous efforts in research and development of this new class of wireless networked embedded systems. Despite the significant effort made, successful deployments and real-world applications of sensor networks are still scarce, labor-intensive and often cumbersome to achieve. In this article, we survey prominent examples of sensor network deployments, in particular for environmental monitoring applications, their interaction with the real world and classify a number of potential causes for errors and common pitfalls. In the second half of this work, we present methods and tools to be used to detect failures, identify and understand root causes. These instrumentation techniques and analysis tools are specifically designed or adapted for the analysis of distributed networked embedded systems at the level of components, sensor nodes, and networks of nodes.

Poster abstract: BTnodes -- a distributed platform for sensor nodes

We motivate a prototyping platform for ad hoc networking research showing some requirements and constraints. The architecture of the BTnodes, each of which can store information, compute and communicate, is explained in conjunction with some demo applications that have been implemented. Important requirements and design trade-offs to be able to support multiple, compatible communication interfaces, to handle limited resources, for power-aware operation and for efficient testbed deployment are discussed.

Practical time synchronization for Bluetooth Scatternets

By means of so-called Scatternets, Bluetooth provides the ability to construct robust wireless multi-hop networks. In this paper we propose a practical protocol for time synchronization of such Bluetooth multi-hop networks. Our protocol makes use of the internal clock maintained by Bluetooth, requires minimal communication overhead, and provides an accuracy of few milliseconds across multiple hops. The protocol has been implemented and evaluated on BTnodes, an embedded computing platform which uses Bluetooth for ad hoc networking.

BurstMAC — An efficient MAC protocol for correlated traffic bursts

Many sensor network applications feature correlated traffic bursts: after a period of idle time with almost no network traffic, many nodes have to transmit large amounts of data simultaneously. One example is volcano monitoring, where rare eruptions trigger many nodes simultaneously to transmit seismic data traces. For such applications, a MAC protocol should exhibit a low overhead both in idle mode and during correlated traffic bursts. Existing MAC protocols only meet one of these requirements, but not both at the same time. Contention-based protocols such as SCP-MAP have very low overhead in idle situations, but exhibit high overhead during correlated traffic bursts due to contention. Scheduled protocols such as LMAC avoid contention overhead during correlated bursts, but exhibit a significant overhead in idle mode due to control traffic. We propose BurstMAC, a new MAC protocol specifically designed for applications with correlated traffic bursts. BurstMACs idle overhead is almost as low as that of SCP-MAC and at the same time provides better throughput and lower energy overhead than LMAC during correlated traffic bursts.

Increasing the visibility of sensor networks with passive distributed assertions

When deployed in a real-world setting, many sensor networks fail to meet application requirements even though they have been tested in the lab prior to deployment. Hence, concepts and tools for inspection are needed to identify failure causes in situ on the deployment site. Tools for inspection should minimize the interference with the sensor network to, firstly, ensure that failures of the sensor network do not break the inspection mechanism, and, secondly, to ensure that the inspection mechanism does not change the behavior of the sensor network. In this position paper, we propose passive distributed assertions (PDA) as a novel tool for identifying failure causes. PDA allow a programmer to assert certain predicates over distributed node states. Packet sniffing is used to detect failed assertions, thus minimizing the interference with the sensor networks.