A wireless sensor network (WSN), as exemplified in Figure 1, is a self-organized, often multi-hop, wireless network, formed by a possibly heterogeneous composition of sensor nodes, which are spread out over an area of interest. These nodes are small embedded devices, able to gather various information about their surroundings, such as temperature or vibration. They are constrained in many ways (e.g., processor, memory), but energy is generally considered to be the scarcest resource, due to limited battery capacities and inefficient energy harvesting. Moreover, as such networks are often deployed in hostile or remote areas, replacing batteries may be infeasible. WSNs commonly follow a data gathering paradigm, in which all collected data is forwarded to a sink. The sink may then store the data for later retrieval, perform further computation on it, or relay it directly to a server for archiving and analysis. As a sensor node’s radio is usually its most power hungry component, efficient routing of data messages from their point of creation to the sink plays a vital role in energy conservation and is one of our main research interests.
For the current version of SensorScope , in order to avoid load-balancing problems and the maintenance of a routing tree, such as the one illustrated in Figure 1b, we have decided to let nodes choose their next hop from a self-determined set of possible relays each time a packet has to be sent. The choice is made at random, favoring, however, better neighbors (from an estimated link quality point of view). While this scheme, named EPFLnet routing, guarantees that all data messages will eventually reach the sink and may also result in alleviated load-balancing, it is certainly not energy-optimal.
In fact, in most wireless networks with unreliable links, the probability of a packet being successfully received by at least one node in a set of neighbors is greater than the probability of one specific node receiving it. This observation motivates the idea of opportunistic routing (OR). In OR, the next-hop routing decision is finalized only after a packet has been sent, allowing the network to opportunistically take advantage of its broadcast nature and outcomes that are inherently random and unpredictable. As WSN topologies commonly possess many more links than those of a minimally spanning tree (cf. Figure 1a and 1b), this inherent diversity should be exploited when sending data from the sensor nodes over multiple hops to the sink. A low-overhead coordination protocol is, however, needed for the actual receivers of a packet to agree upon which of them should further forward it. Although the problem of choosing the optimal forwarding set and that of receiver coordination are often treated disjointly, they are indeed related, as a larger set of forwarders will most-likely incur a higher coordination cost. We are currently investigating possible improvements of an integrated solution in three areas:
- network reliability,
- energy consumption, and
Figure 1: Prototypical topology of a wireless sensor network (a), together with the possible layout of a routing tree (b).
G. Barrenetxea, F. Ingelrest, G. Schaefer, M. Vetterli, O. Couach and M. Parlange, SensorScope: Out-of-the-Box Environmental Monitoring, The 7th ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN 2008), pp. 332-343, 2008.
[detailed record] [bibtex]
G. Barrenetxea, F. Ingelrest, G. Schaefer and M. Vetterli, Wireless Sensor Networks for Environmental Monitoring: The SensorScope Experience, The 20th IEEE International Zurich Seminar on Communications (IZS 2008), 2008.
[detailed record] [bibtex]
Project Period, Funding Source:
September 2006 – ongoing,