Section: New Results

Routing in FUN

Participants : Nicolas Gouvy, Xu Li, Nathalie Mitton.

Wireless sensor and actuator/robot networks need some routing mechanisms to ensure that data travel the network to the sink with some guarantees. The FUN research group has investigated different geographic routing paradigms. It first has considered a static network in which the routing either enhances the energy cost [22] , [10] , balances the load over nodes [21] , [8] or respects traffic priorities [18] .

A more complex routing paradigm has been proposed in [25] for $k$-anycasting. In $k$-anycasting, a sensor wants to report event information to any $k$ sinks in the network. This is important to gain in reliability and efficiency in wireless sensor and actor networks. In this paper, we describe KanGuRou, the first position-based energy efficient $k$-anycast routing which guarantees the packet delivery to $k$ sinks as long as the connected component that contains $s$ also contains sufficient number of sinks. A node $s$ running KanGuRou first computes a tree including $k$ sinks among the $M$ available ones, with weight as low as possible. If this tree has $m\ge 1$ edges originated at node $s$, $s$ duplicates the message $m$ times and runs $m$ times KanGuRou over a subset of defined sinks. Simulation results show that KanGuRou allows up to $62\%$ of energy saving compared to plain anycasting.

We then assumed that the sink that collects data is actually mobile and travels the network. Sensor nodes need thus to update the position of the sink in a smart fashion in order to limit the overhead generated by this update. In [9] , we propose a novel localized Integrated Location Service and Routing (ILSR) scheme, based on the geographic routing protocol GFG, for data communications from sensors to a mobile sink in wireless sensor networks. The objective is to enable each sensor to maintain a slow-varying routing next hop to the sink rather than the precise knowledge of quick-varying sink position. In ILSR, sink updates location to neighboring sensors after or before a link breaks and whenever a link creation is observed. Location update relies on flooding, restricted within necessary area, where sensors experience (next hop) change in GFG routing to the sink. Dedicated location update message is additionally routed to selected nodes for prevention of routing failure. Considering both unpredictable and predictable (controllable) sink mobility, we present two versions. We prove that both of them guarantee delivery in a connected network modeled as unit disk graph. ILSR is the first localized protocol that has this property. We further propose to reduce message cost, without jeopardizing this property, by dynamically controlling the level of location update. A few add-on techniques are as well suggested to enhance the algorithm performance. We compare ILSR with an existing competing algorithm through simulation. It is observed that ILSR generates routes close to shortest paths at dramatically lower (90% lower) message cost.

When the network is composed of mobile sensors that have the faculty to control their mobility, this property can be exploited to enhance routing performance. In [3] , we are interested in energy-aware routing algorithms that explicitly take advantage of node mobility to improve energy consumption of computed paths. Mobility is a two-sword edge however. Moving a node may render the network disconnected and results in early termination of information delivery. To mitigate these problems, we propose a family of routing algorithm called CoMNet (Connectivity preservation Mobile routing protocols for actuator and sensor NETworks), that uses local information and modifies the network topology to support resource efficient transmissions. Our extensive simulations show that CoMNet has high energetic performance improvement compared to existing routing algorithms. More importantly, we show that CoMNet guarantees network connectivity and efficient resource consumption.