Scheduling algorithm based on sender buffer backlog for real-time application in mobile packet networks

Abstract

The demand for high-speed data communication and the use of real-time applications in mobile networks are becoming increasingly high, and communication systems that specialize in downlink packet transfer are being developed and deployed. In these high-speed downlink packet networks, scheduling at the base station is one of the key technologies to accomplish such high-speed. A scheduling algorithm that is based on the proportional fair criterion is widely used in commercial networks, but the problem with such algorithm is that it does not schedule real-time application's traffic efficiently. This paper proposes a new scheduling algorithm for real-time applications that does not need substantial change in the existing system and coexists efficiently with the proportional fair criterion. We evaluate the characteristics of the proposed algorithm through computer simulation and show that the efficiency of the proposed algorithm is better than that of existing algorithms.

Mitigation of the Effects of Network Outage on Video QoE Using a Sender Buffer

“…In Reference [31], the authors investigate the double-buffer traffic shaper, implemented using Token and Leaky Bucket techniques, to adjust the video frame rate inflow into the TCP sender buffer of a multimedia application source across a slow-speed link. The authors of [32] introduced a scheduling algorithm based on the sender buffer backlog for real-time application. It schedules time slots based on the sender buffer backlog at the base station, which is believed to be correlated with a play-out buffer backlog at the receiver.…”

Scheduling algorithms for 4G wireless networks

“…The main drawback of the PF scheduling stems from the use of a ratedependent-only utility function that makes the system unable to ensure traffic delay requirements. Several proposals try to solve this limitation [27,28] maintaining the PF utility function for best-effort traffic and defining new utility functions for delay sensitive traffic flows. In the DS-PF scheduler {Delay Sensitivity PF) [27] real-time sessions increase their priority when a certain delay value is trespassed, while in the SB-PF {Sender Buffer PF) [28] proposal the objective is to avoid playout buffer starvation.…”

“…Unfortunately, in these mixed cases with different utility functions, the aggregate system utility optimization is much more complex. To preserve PF simplicity, both Barriac in [27] and Koto in [28] propose sub-optimum optimization algorithms.…”