Retry Limit Based ULP for Scalable Video Transmission over IEEE 802.11e WLANs

Fast retry limit adaptation for video distortion/delay control in IEEE 802.11e distributed networks

“…Accordingly, several retry limit adaptation methods have been proposed in the research literature [5][6][7][8][9][10][11][12][13][14][15][16]. The QoS strategy presented in [5] adopts a priority queueing also at the network layer, in order to relate the adaptation to the tradeoff between drop probability and buffer overflow rate.…”

“…In [7] the conventional count-based retransmission scheme is replaced by a time-based one, in which the deadline is determined by the expiration time and the importance of the intercoded frames. A retry limit adaptation method for scalable videos is developed in [8] by considering the collision probability as a load indicator. The reciprocal influence among the nodes and the ACs on the selection of the retry limit is analyzed in [9], where an adaptive algorithm is derived from the numerical solution A C C E P T E D M A N U S C R I P T of a nonlinear system.…”

“…In particular, S can support four ACs, which are numbered according to q = 1 (VO), q = 2 (VI), q = 3 (BE), q = 4 (BK), thus indicating that a lower q value identifies a higher priority. Assume that each AC of each source remains nonempty once a packet is successfully transmitted, hence considering, as in [8,10], saturated traffic conditions. For the BE and BK ACs, the saturation assumption is widely accepted, since it derives from usual file transfer applications.…”

“…To further test the proposed solution, the obtained results are compared to those achievable using the retry limit adaptation algorithm presented in [8]. This algorithm relies on an unequal loss protection approach developed according to the collision probability and to a strategy of differentiation of the packets into groups, in which the most important packets are associated to higher retry limits.…”

“…This algorithm relies on an unequal loss protection approach developed according to the collision probability and to a strategy of differentiation of the packets into groups, in which the most important packets are associated to higher retry limits. The solution in [8] has been selected for its very low computational cost, thus making significant the comparison with the algorithm proposed in this paper.…”