Igor Talzi

PermaSense: investigating permafrost with a WSN in the Swiss Alps

Currently, there is a lack of stand-alone geo-monitoring systems for harsh environments that are easy to configure, deploy and manage, while at the same time adhering to science grade quality requirements. In a joint computer and geoscience project we have built and deployed a wireless sensor network for measuring permafrost related parameters. Using these high-precision data, geo-scientists will be able to calibrate their heat flux models in order to better predict the stability of steep rock slopes in the alps. In this paper we describe our system from a computer science and system point of view and report on some lessons learned, especially in the domain of sensor design, power-awareness and reliable data flow.

Providing data integrity in intermittently connected Wireless Sensor Networks

Intermittently connected wireless sensor networks (WSN) for scientific environmental monitoring raise the issue for reliable data gathering which is a key factor for having a consistent data stream as a result. First, measurements must be taken and securely saved on a node. Per node local repository for measurements is a basic mechanism for making possible to retransmit any block of data towards a data sink if it has been lost or damaged. Second, high reliable data communication channels are established between nodes, including MAC-layer and application-layer ldquoend-to-endrdquo acknowledgments. The last, and the most important aspect is using a multi-level data buffering mechanism based on an ldquoagingrdquo-factor for reducing power consumption taken by radio transmissions. We describe our solution, combining all these techniques, which is intended for use in a permafrost monitoring project. In our case where half-/hourly measurements are collected over a whole year and nodes can become disconnected for several months, we still manage to provide loss-free and power efficient data delivery.

Force-based navigation in wireless sensornets

Certain scenario-oriented network tasks require a non-traditional approach to routing. This happens when routing should become aware of the nodes state defined beyond the network layer (e.g., sensed value, current physical location, etc.). In this paper we present an approach that helps software elements to navigate through the network based on the metaphors of fields and forces known from physics. The elements encapsulate logic to make a decision on the next move in the dynamic environment that is formed by overlapping, interacting fields and represent a physical value (e.g., fire) or a role (e.g., sink node). As a result, the force is induced on a node which pushes the element in a certain direction, towards or even away from the field source. We illustrate the feasibility of our solution using an emergency evacuation scenario for Wireless Sensor Networks (WSN). We show that our system is simple yet provides more flexibility and functionality than existing counterparts.

Online code compression in Wireless Sensor Networks

A problem in Wireless Sensor Networks (WSN), an extremely resource-limited system, is that re-tasking by replacing complete or partial code images is both disruptive and energy intensive. In this paper we report on a dynamic code compression scheme for mobile code that we implemented for a WSN. We use a fine-grained code mobility scheme based on capsules, ChameleonVM, where network functionality can be deployed on demand and for several tasks in parallel. Within a tasks virtual network segment we let our ChameleonVM optimize the assignment of instruction bits: Mobile code instructions are re-coded on a regional basis and depending on the actual code sequences used, incrementally leading to smaller capsules. We demonstrate the operation of our online compression scheme with a time-sync protocol and discuss its performance.