No abstract

Delay Tolerant Networks are emerging as a new form of network in which sending and receiving nodes may not be reliably connected to each other in a traditional sense but must instead rely on mobile nodes to ferry messages through the network. Applications operating in these dynamic mixed environments would like their connections to be supported by the network technology best suited to the combination of the communication session's requirements and instantaneous network context. In this paper, we explore the systems issues related to enabling such a symbiotic network architecture. We develop an adaptive middleware that enables connections to seamlessly migrate from one communication style to another in response to changing network or application conditions. We delineate the properties of the middleware architecture, describe an initial implementation, and provide a simulation analysis that demonstrates that, given perfect context-awareness, significant performance improvements can be made by using such an adaptive middleware.

No abstract

Despite increasingly realistic vehicular network simulations, the effects of real-world mobility on network and application performance in vehicular networks are still not well understood. We present Pharos, a small-scale vehicular network testbed with "push-button" experiment repeatability and develop a framework for analyzing network performance of vehicular networks simultaneously in simulation and in the real world. We empirically study the differences between real-world and simulated connectivity. Early experiment results using our vehicular testbed show significant differences between simulated and actual movements resulting in differences in wireless connectivity. Because of this, we implement a trace mobility model that allows the OMNeT++ simulator replay actual GPS-based movement traces collected by the testbed and scale to larger networks.

As computing becomes increasingly mobile, the demand for information about the environment, or context, becomes of significant importance. Applications must adapt to the changing environment, however acquiring the necessary context information can be very expensive because collecting it usually requires communication among devices. We explore collecting reasonably accurate context information passively by defining passively sensed context through network overhearing, defining context metrics without added communication cost. We use this framework to build a small suite of commonly used context metrics and evaluate the quality with which they can reflect ground truth using simulation. We also provide an implementation of this passive sensing framework for Linux, which we evaluate on a real mobile ad-hoc network using mobile autonomous robots with commodity 802.11 b/g wireless cards.velocity information. The extra network traffic these mechanisms generate places an increased burden on the already taxed network, making it difficult to justify the use of context-awareness in the common case. If the overhead of sensing context information can be reduced, the benefit of the availability of the information is increased.We define a framework for defining passively sensed context metrics based on network eavesdropping (Section 3). Our approach focuses on sensing context with zero additional communication overhead. Our context metrics do not provide the exact measure of context that their active counterparts may provide, but we demonstrate the measures' fidelities match traditional measures of context. We use this framework to create instantiations of three common network context measures (Section 4). For each of these metrics, we evaluate the specificity of the passively sensed context metric with respect to the ground truth (Section 5). Our work shows that passive sensing of network context can inexpensively provide information about the state of the world and that, especially when these metrics are correlated with each other, enable adaptive applications in environments where traditional active context sensing is cumbersome. Related WorkThe demand for adaptive mobile applications indicates the need for efficient contextawareness. Much work has focused on supporting software engineering needs through frameworks and middleware that provide programming abstractions for acquiring and responding to context. For example, Hydrogen [10] defines a completely decentralized architecture for managing and serving context information. Hydrogen's abstractions are unconcerned with how context is sensed; clearly, performing context acquisition efficiently is important to the success of such a framework. Many other projects have also looked at reducing the cost of context sensing. Several of these take an applicationoriented perspective, identifying what high-level information the application desires and only acquiring information necessary to support an application's desired fidelity [26]. SeeMon [13] reduces the cost of context by onl...