Programmable wireless devices which can perceive the current radio environment, decide available spectra, and dynamically change the radio access method, and networking protocols have been proposed to improve spectrum usage, interference mitigation, and connectivity. However, these transitions are currently lacking at upper layers. In this paper, we propose an upper layer cognitive system beyond channel sensing effectiveness and spectrum utilization achieved in adjusting PHY and MAC parameter settings. The proposed cognitive video streaming system achieves end-to-end goals at an upper video layer: quality of experience, service continuity, and survivability. Based on the link state of the receiver, the proposed system identifies areas of importance in the image stream, allocates higher bandwidth for the important areas compared to the other areas which is adjusted to use the less bandwidth with CoSA encoder and decoder, and receive the link state feedback from the receiver. To further understand the CoSA's benefit, we implemented two real test-beds for the cognitive video streaming system where the one uses conventional IEEE 802.11 networking technologies and the other employs a software defined radio platform, and performed a performance evaluation study in order to see the effectiveness of the proposed scheme. The results indicate that the proposed system can dynamically change actual data rates according to SINR feedback, which results in at least 180% improvement of transmission time compared to conventional method, while maintaining PSNR range between 30 to 40 dB with eminently reduced data size.