No abstract

No abstract

On-line video services are becoming a large part of the daily routines of people all over the world, where most of the content is accessed through wireless networks. Therefore, it is of ever growing importance that the negative aspects of these types of error prone networks are lessened in order to ensure adequate quality of the delivered video streams. Forward Error Correction (FEC) techniques allow the stream to be protected with an amount of redundancy to preserve the video quality during transmission. Nevertheless, some FEC schemes do not make an efficient usage of the available network resources due to unnecessary use of redundancy as a result of video-unawareness. The adaptive FEC mechanism proposed in this paper uses the motion intensity characteristics of the video and the network loss state to deliver the video streaming with adequate Quality of Experience (QoE), while keeping the use of network resources to a minimum level. It does so from a combined use of a Random Neural Network (RNN) for motion intensity classification and an Ant Colony Optimization (ACO) scheme for dynamic redundancy allocation. QoE metrics are used to assess the performance of the mechanism showing its advantages over adaptive and nonadaptive protection schemes.

The video delivery over wireless networks has risen in popularity in the recent years. However, in order to provide a high quality of experience (QoE) to the end users, it is necessary to deal with several challenges ranging from the fluctuating bandwidth and scarce resources to the high error rates. The use of these error-prone networks unveils the need for an adaptive mechanism to ensure the quality of the delivered video streams. Adaptive forward error correction (FEC) techniques with QoE assurance are desired to protect the stream, preserving the video quality. The adaptive FEC-based mechanism proposed in this article uses several video characteristics and packet loss rate prediction to shield real-time video transmission over static wireless mesh networks, improving both user experience and the usage of resources. This is possible through a combination of a random neural network, to categorise motion intensity of the videos, and an ant colony optimisation scheme, for dynamic redundancy allocation. The benefits and drawbacks are demonstrated through simulations and assessed with QoE metrics, showing that the proposed mechanism outperforms both adaptive and non-adaptive schemes.

No abstract