Delay minimization with channel-adaptive packetization policy for random data traffic

Packet lengths have crucial impact on the end-to-end delay in a queuing system over noisy time-varying wireless channels that is overlooked in most system designs. In this paper, a delay optimal adaptive packetization policy in network layer is proposed that adjusts framing intervals based on the underlying physical layer parameters in order to minimize the end-to-end packet delivery time. The arrival stream of fixed-length symbols with exponential inter-arrival time distribution are bundled into packets in an optimal manner and are scheduled for transmission over a noisy wireless channel with Automatic Repeat Request (ARQ) retransmission mechanism. The underlying channel quality in terms of BER is considered known to the scheduler. The proposed solution not only minimizes the end-to-end latency but also prohibits potential queue instability in dynamic channel situations.