| Popis: |
We propose two multipath rate control algorithms that guarantee bounded end-to-end delay in multihop wireless networks. Our work extends the previous research on optimal rate control and scheduling in multihop wireless networks, to support inelastic delay requirements. Using the relationship between dual variables and packet delay, we develop two alternative solutions that are independent from any queuing model assumption, contrary to the previous research. In the first solution, we derive lower bounds on source rates that achieve the required delay bounds. We then develop a distributed algorithm comprising scheduling and rate control functions, which requires each source to primarily check the feasibility of its QoS before initiating its session. In the second solution we eliminate the admission control phase by developing an algorithm that converges to the utility function weights that ensure the required delay bounds for all flows. Both solutions carry out scheduling at slower timescale than rate control, and consequently are more efficient than previous cross-layer algorithms. We show through numerical examples that even when there are no delay constraints, the proposed algorithms significantly reduce the delay compared to the previous solutions. |
