The resource constraints of mobile devices put strict limits on the utility and quality of wireless applications in an increasingly nomadic world. While there has been an abundance of previous work on resource management, there has been a dearth of research to address the complex problem of resource interdependencies for mobile systems. For example, most modern mobile devices employ multiple techniques to reduce their energy requirements, however, these techniques focus on individual resources and are unaware of the effects their deployment may have on other resources and application performance. Further, in distributed systems, devices typically are selfish in the sense that they manage their own resources without considering the consequences for resources and applications on remote devices. This NSF CAREER research project is developing both, an experimental framework for investigating these interdependencies and design issues, and new service concepts to provide resource management solutions that enable the efficient integration of multiple constraints, resources, and devices. The new approach, dubbed judicious resource management, moves away from traditional, isolated resource management solutions and focuses on the prevention of unanticipated side effects of interdependent resource adaptations. The broader impacts of this project include an experimental setting that should serve as catalyst for future research on distributed resource management, paving the way for new classes of protocols, management techniques, and approaches to integrate them. This work closely integrates research and education activities, using a single experimental test bed for research, education, and outreach efforts.
Summary of Project Outcomes Over the last couple of years, our society has grown increasingly reliant on mobile technology such as smartphones and tablets for our communication, networking, entertainment, and many other activities. However, if wireless infrastructure (such as cell towers or base stations) fails, these mobile devices are often rendered useless. In this project, the goal was to develop the tools required to establish direct device-to-device communication when wireless infrastructure is unavailable, destroyed, or of poor quality (e.g., after a disaster). The resulting SPIRIT framework allows mobile devices to discover and connect to other mobile devices in their proximity, reliably propagate data between SPIRIT peers, even over multiple hops. Using these "multi-hop" communications, it is possible for devices to share data even if they are not in direct reach. The key outcomes of the SPIRIT framework are its robustness and resource efficiency features. Robustness. SPIRIT peers can maintain communications even when all devices are mobile. It does so by employing a robust publish-subscribe approach, where a publisher (a data source) and its subscribers (one or more data sinks) maintain connectivity over multiple hops using frequent location updates exchanged between each other. These updates are exchanged using a geographic forwarding protocol, where the update frequency depends on the mobility (speed, direction) of the devices. The key to this technique is the Overlay Multicast Data Distribution Tree (OMDDT) formed by SPIRIT, which, compared to other common data routing approaches, leads to an increased data delivery success rate by 10-18% depending on the device mobility. Resource Efficiency. SPIRIT uses a modified version of the Euclidan Steiner Tree heuristic to obtain a data dissemination tree structure between a publisher and its subscribers. In this approach, multiple receivers are grouped together and initially messages are propagated towards these groups of receiver. When a message approaches a group, the OMDDT algorithm may decide to create duplicates of a message to be sent to sub groups of the original group. Finally, once a message approaches the predicted region of a device (which changes over time due to mobility), the message is broadcasted to all devices in the vicinity to increase the likelihood of reaching the intended receiver. This approach combines the goal of reliably reaching message receivers, while minimizing the number of messages and message transmissions required for communicating with all receivers. Application Scenario. A third contribution of this project is the design and implementation of a mobile application for smartphone devices that builds on the SPIRIT infrastructure to help first responders in emergency situations. This application allows paramedics to monitor the breathing activities of multiple patients simultaneously by placing sensors (e.g., accelerometers) on their chests, which then report the breathing reports (and any abnormal events) to one or more paramedics. The goal of this part of the project is to show the feasibility of using the SPIRIT architectures for real-world scenarios where direct device-to-device communication may be required.
