Abstract

The main scenario is the one of a naturalistic park in which a network of wireless sensors is deployed with a main duty of environmental monitoring. Sensors are in part fixed and in part mobile: they collect specific data, partially elaborate them, send them to sinks for centralized and/or constraint- less elaboration, collaborate to disseminate information to interested nodes which can be data collectors or actuators. Such a network is actually a wireless sensor and actor network which topology implies a non trivial routing: nodes can be mobile or fixed; sensor or actuator or both; some can be strictly constrained on buffer size, power supply, radio range and computational power while other not. This high heterogeneity traduces in a complex network. In this first year, I am currently focusing on the fixed node subnetwork. These nodes are spread on a broad area, only few of them can be wired to an external network, many are likely to be physically accessed by staff very few times over a long period. This means that memory and power management are main concerns during nodes workout. One of the main way to save battery power is to switch the wireless interface off when the communication is not needed as, for these devices, the network communication is by orders of magnitude more expensive than the other local activities.

Considering that each node has this kind of duty-cycle, the fixed nodes subnetwork results to be intermittently connected. We assume that a network manager assigns a specific duty-cycle to sensor nodes, which is disseminated in a specific startup phase. Then sensors start to exchange their duty cycle and construct their future communication behavior on the duty cycling information propagated by the neighbours. We construct a decentralized time-aware delay tolerant routing protocol which aims to provide data delivery to specific points and minimize end-to-end cost, considering that no contemporary end-to-end path could exist between a source-destination node pair. I am currently verifying the algorithm correctness and running it by means of simulations to identify its basic properties.