Research On Resource Management And Service Performance Optimization Of Smart TV Terminal

With the advancement of triple play, the cable networks of our country have basically achieved integration on the network physical layer with telecommunications networks and the internet, and the integration on the service application layer becomes the key point of present stage development. The development of the integration services need to rely on the support of the integration terminals, but various manufacturers of the traditional digital terminals have different hardware and software protocols, which restricts the development of integration services. Smart TV which is equipped with the operating system and open platform becomes the product of integration terminal fields under the background of triple play.Our work in this dissertation is based on the Program "the development of integration network scope" and the Program "new generation of television service system supporting cross-regional and multi-service-providers". This dissertation is concerned with the resource management and service performance optimization of Smart TV terminal. A highly effective resource allocation algorithm is utilized to optimize the allocation of system resource on the system layer for the sake of guaranteeing the QoS of applications; and in order to improve the service performance of video streaming application, video traffic smoothing and adaptive media playout technologies are adopted to optimize the application logic design on the service layer. The main contributions are given as follows:1) Due to the limited nature of hardware resources in embedded system and unreasonable resource allocation, Smart TV cannot guarantee the QoS of applications. In order to solve this problem, a resource management model, which combines with user application preference and application QoS contract, is proposed. Then a fast heuristic algorithm based on constructing convex hulls of working points, named WPCHRA, is presented for this model. The penalty vector is used to transform multi-dimensional resources into the one-dimensional compound resource, and the segments of the convex hull up-boundary are adopted to guide the selection of the working point. This algorithm can allocate system resource reasonably, guarantee the QoS level of the preference application and improve the user experience. Compared with the heuristic algorithm M_HEU with a time complexity of O(mn2L2), the WPCHRA algorithm whose time complexity is only O(nLm+nL log nL), can obtain near-optimal solution in a very short time, hence is suitable for Smart TV operating system real-time resource allocation.2) High burstiness and fluctuation of variable rate videos (VBR) during the transmission process makes the management of the video stream admission control and network bandwidth more complex. In order to solve this problem, a traffic smoothing algorithm based on the judgment of buffer filling degree, named BFDAS, is presented. Given the buffer capacity and the start-up delay, the feasible transmission scheme could be obtained at the first running of the BFDAS algorithm. Compared with the fixed segmented algorithm PCRTT, the video traffic of BFDAS algorithm reduces peak rate by12%, rate change numbers by76%and the ratio of rate change by15%, so the data transmission is more stable. Because of the half average buffer utilization, it could also better balance the high efficiency utilization of the buffer with the need of supporting VCR operations for clients. Furthermore, the fixed start-up delay characteristic of BFDAS algorithm makes it more suitable in the practical service systems.3) The actual network channel condition is dynamic and the delay of VBR video data is stochastically time-varying. Large delay may result in buffer underflow, even playout interruptions for client. In order to solve this problem, an adaptive media playout algorithm based on the estimation of frame rate, named FREAMP, is proposed. The amount of data occupancy and the frame arrive rate are taken in to consideration in the FREAMP algorithm to decide the playout rate of the video. When the amount of data occupancy is in the warning region, the playout rate will slow down or speed up quadratically. While the amount of data occupancy is in the safe region, the playback speed is determined based on the estimation frame arrival rate. The purpose of FREAMP algorithm is to stabilize the data occupancy of buffer as far as possible, reduce buffer underflow probabilities and then increase the continuity of video playout. Experimental results show that in the same network channel state, compared with AMP-Live and AMP-Gradual algorithms, the playout interruptions of FREAMP were reduced by53.2%and33.7%respectively, and the playout delay was reduced by12.7%and8.9%respectively.