Q16Mobile Computing

Data dissemination in mobile computing refers to the delivery of data from a set of data servers to potentially very large numbers of mobile clients over a bandwidth-constrained, error-prone wireless channel. Because mobile devices are limited by battery power, intermittent connectivity, and scarce uplink bandwidth, dissemination strategies are designed to minimize how often clients need to actively request data and how long their radios must stay powered on to receive it.

• Pull-based: The traditional client-server model, where a client explicitly sends a request and the server responds, as in web browsing. This scales poorly as the number of clients grows, since the server must handle each request individually.
• Push-based: The server proactively broadcasts data without waiting for explicit requests, for example continuously transmitting stock prices or weather updates. This scales well because a single broadcast serves an unlimited number of listening clients.
• Hybrid: Combines both, broadcasting popular data proactively while allowing clients to pull less common data on demand, balancing scalability with responsiveness.

Broadcasting is the primary mechanism for push-based dissemination. The server organizes data into a broadcast schedule, sometimes called a broadcast disk, and transmits it repeatedly and cyclically over a shared channel. Clients tune in to the channel and filter out the items they need, without ever transmitting a request, which conserves the client's limited uplink power. Frequently accessed items can be broadcast more often than less popular ones to reduce the average access latency. To further save power, an index can be interleaved with the data so that a client can compute when its desired item will appear, switch to a low-power doze mode, and wake up just in time to receive it, a technique known as indexed broadcasting.