A testbed for performance analysis of 'see-what-I-see' video calls and quality feedback

Todays' peer-to-peer video communications are no longer restricted to 'talking heads' exchanges; they increasingly involve sightseeing and motion scenes. These new forms of video content are bandwidth-demanding and require some guarantee for a satisfactory quality of experience (QoE). However, QoE becomes a challenging issue in presence of network impairment. This paper presents the design and experimentation of a testbed for analyzing the quality of simulated video calls on error-prone channels. Functional blocks are included for in-service media transport measurements and feedback reports, as well as off-line video quality assessment in order to evaluate the QoE perceived by users in different contexts. The system is validated using video content types which simulate 4 different forms of see-what-i-see video calls. Feedback reports are analyzed in order to identify their limits in inferring the end-user video quality. Experimental results are reported that illustrate the use of the testbed exploitation in measuring user perceived quality as a function of network conditions. Finally, the paper gives directives for the network impairment limits that can be tolerated by users.