Network daemons for distributed sensor networks

The reliance on the Internet technologies by DSNs often manifests in severe performance limitations. For example, in wide-area DSNs, the nodes cannot control routes to avoid the congested regions or accumulate bandwidths over multiple paths (without drastically changing the infrastructure) because the routing at the network core is solely determined by the underlying routers, which are exclusively controlled by the different service providers. Furthermore, there is no support for realizing stable channels for control purpose over such wide-area networks. The throughput achieved for a control channel that employs the most widely used transmission control protocol (TCP) typically underflows under high traffic and overflows under low traffic. Moreover, such a TCP-based control channel may experience very complicated end-to-end transport dynamics in time-varying network conditions. In IEEE 802.11 networks for a team of mobile robots, which, for example, are deployed to assess the radiation levels of a remote area, an infrastructure of access points must be set up prior to the operation [3]. Such a requirement is obviously meaningless if the very goal of the robot team is to assess the suitability of the region for human operation. More generally, in DSNs of mobile nodes, the challenges are to form an ad hoc wireless network without the infrastructure [4] and to cope with the dynamic changes in network connectivity. Note that the node movements are treated as aberrations in the current Internet environments, where as such movements are shown to improve the message delivery in mobile ad hoc networks [5].