On-demand multimedia data broadcast in MIMO wireless networks

Given a set of multiple requests from clients equipped with M antennae and a wireless network of m channels, on-demand data broadcast requires to find an optimal schedule of broadcasting (placing) all requested data items of R on a set of channels C as evenly as possible under the constraint that each client may use at most M channels. Such a schedule is important for overcoming the shortcomings of wireless networks such as asymmetric bandwidth between uplink and downlink, and limited battery life of mobile devices. Existing schemes for data broadcast assume single-antenna clients and will result in significant bandwidth wastage and client's data download time increase. To overcome these problems, we propose a novel approach for disseminating multimedia data in a MIMO wireless network by converting it to the multiprocessor scheduling problem where requests and antennae are regarded as tasks and processors respectively. We present three schemes (LFOS, BFOS and BBOS) under this approach: LFOS scheduling data items of largest sizes, BFOS adopting the best matching between data items and channels, and BBOS partitioning data items properly to balance the broadcast cycles of all channels. In comparison with the existing schemes based on single-antenna broadcast, our schemes improve access latency and channel bandwidth usage significantly. This has been verified through extensive experimental results.