If you sit back on a Sunday afternoon, thinking about the possible application areas for wireless sensor networks, the outcome is stunning. Systems can benefit from situational awareness and context sensitivity which can be accomplished by many sensors, smart communication between them and smart reasoning. Our vision is that in the near future wireless sensor networks (WSN) will become an integral part of our environment. Pervasive or ubiquitous technology will play an increasing role in society, e.g. energy efficiency in buildings, monitoring applications in healthcare and support for elderly people (independent living), indoor localization, monitoring in logistic systems, and many more.
However, a number of topics have to be addressed before WSN can play that integrated role. In most applications the network will consist of a very large number of sensor- and actuator nodes. So the nodes have to be low cost, very low power and small form factor. In order to have a manageable network it has to be robust, scalable or scale free, self organizing and it has to adapt itself to the context in which it is used.
MyriaNed, a self organizing, gossiping Wireless Sensor Network
DevLab’s approach is called MyriaNed, which is inspired by biological processes where many nodes (birds, ants, cells) operate in large distributed systems (resp. flocks, organized colonies, organisms). It is a bottom up approach, where the behavior of a single element (node) will result in emerging behavior of the system (application).
An arbitrary node will broadcast a new message into the network. In general this will be initiated by a new sensor reading. This message will reach a number of nodes, but not all. Because the nodes also re-sent the received information, the information is spread through the network, and eventually the message is “known” by every node in the network. Epidemic message flooding through the network called gossiping. Every node will have this ‘behavior’, so at any time, every node is familiar with the state of all other nodes in the network, hence the state of the (distributed) system. We call that shared state, which reflects the environmental situation that the network is in. So the system is situational aware and is able to adapt itself to the context. Every node will play a role in that environment in many possible ways: act on sensor information, combine information from multiple sensors (sensor fusion), recognize patterns and trends, perform high level reasoning, calculate aggregated data and eventually driving actuators or interact with the user interface.
The communication protocol, which is event driven and based on an energy efficient TDMA schedule, is extended with the gossiping and shared state layers in such a way that the number of nodes is not limited (scale free). Automatic synchronization with the networks heartbeat and dynamic slot allocation in the TDMA schedule enables nodes to join and leave the network while the network (application) is running. This concept has proven to be very robust and forms the basis of many applications that can be build on top of this.