|
Pocket Switched Networks are mobile networks taking benefits of all data transport opportunities, including the mobility of the users themselves. The name is a play on "packet switched networks", referring to one of the fundamental design choices behind the Internet architecture. With the multiplication of radio-enabled devices carried by humans - literally, in their pockets - this new type of communication is quickly becoming a tangible alternative to infrastructure-based mobile communication, whenever the application can tolerate some delay.
This topic is at the crossroads of several networking research themes. These networks may be considered along the same line as Mobile Ad-hoc Networks (i.e. networks where every node participates as a source, a destination, and a router), with the key difference that no contemporaneous route can be constructed, as pocket switched networks are essentially disconnected most of the time. Delay Tolerant Networks have also been introduced with similar applications in mind, whenever the dynamic Publications & Reports follow, more or less, a schedule. Lastly, one may think of a pocket switched networks as the most extreme generalization of peer-to-peer transfer into the mobile worlds.
Our research emphasizes the characterization of human mobility (in either work-related or unspecified environment) and the condition that makes such opportunistic transport feasible. Several recent advances in the understanding of large complex systems, such as sociological acquaintance networks, make us believe that routing among mobile people could be achieved with practical forwarding algorithms within reasonable memory/bandwidth cost. Today our research agenda includes:
- Human mobility experiments to study the properties of contact process between humans.
- Identify the communities users belong to and how it impacts when and how they meets other people from their groups.
- Characterize the performance of routing schemes.
- Provide network assistance at minimum cost whenever users mobility is not enough.
- Encourage collaboration among users, design incentive to make sure it is maintained.
In other words, in a few years, if we succeed, the best person to send the file you need at any time may very well be one of your neighbours.
|