Several models exist for the definition of Dv (e.g.,
see [15]). Recent results [16] [17] demonstrated that acknowledgment-based transmission
of scalable video under a strict distortion-reduction prioritization of the video packets leads
to an additive distortion-reduction model at the receiver side under packet losses in the
multi-hop network. This additive model is codec-specific and typically expresses the
expected mean square error (MSE) reduction at the video decoder instead of the visual
distortion reduction, since the latter is harder to quantify. See [15] for an example and [18]
for further details on linking the distortion-reduction estimates (Dv) with the packetization
process at the application layer. We assume that an optimized packet scheduling is
performed at the application layer, where all packets v with the same delay deadline
deadline
v d are ordered at the encoder (sender) side according to their expected distortion
reduction Dv [15] [3]. The delay deadline deadline
v d will also be considered as a parameter in
the proposed optimization strategy and it will be defined based on the application
requirements.
At each link in the mesh wireless network, each video flow is subjected to a queuing
delay6, which depends on: (a) the MSDUs from a particular flow (user) that are scheduled
for transmission via the link of interest at the moment when MSDU v arrives; (b) the queue
output rate.
Pages:
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318