• Journal of Korea Multimedia Society
• /
• v.1 no.1
• /
• pp.67-79
• /
• 1998

In this paper, we present the design and analyze the transmission rate of a control algorithm for the client buffer on the VOD systems. The design is based on the server's data transmission rate and clients' data consumption rate. The proposed algorithm stabilizes the client's buffer by reducing the oscillation phenomena of the buffer. And it uses the BCT (Buffer Check Time) to reduce the scheduling load of the client system. The client's data consumption rate and the buffer size are calculated on the basis of BCT. In case that the predicted buffer size operates in the overrun or starvation the buffer. As the load fluctuation of the network, the transmission rate control policy cannot stabilize the client's buffer alone. The media scaling policy, therefore, also performs its complementary part in stabilizing the client's buffer, and performance evaluation of the proposed algorithm is provided through simulation.

• The KIPS Transactions:PartA
• /
• v.10A no.6
• /
• pp.677-684
• /
• 2003

Most policies of VOD services are mainly studied about server system, and make no touch on clients buffer stabilization which is the basis of guaranteeing playback quality and playing without cut off. In this paper, we proposed a skipping frame transfer policy that can sending dynamic buffer control signal to server in accordance with clients buffer state of starvation or overrun, and server transfers a skipped frame to client after receive the control signal. And through a simulation, we show the suggested policy is more efficiency on playback quality by buffer stabilization which is by preventing loss and miss for receiving frame due to grow worse client buffer state.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.11
• /
• pp.34-39
• /
• 2009

In this study, we have proposed the random function arbitration method using a random function manager in other to use the merits and overcome the demerits of fixed priority and round-robin, Starvation phenomenon can be generated from the master having low priority in fixed priority method. Also, round-robin method can't give a priority to important master, however, as the proposed random function arbitration method can be random set the rate of fixed priority and round-robin, we can solve the problems of conventional arbitration methods. From TLM simulation, we confirmed that the bus utilization of master could be controled from 32% to 16% and the starvation phenomenon of master could be prevented.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.3
• /
• pp.69-74
• /
• 2009

We propose the novel Hybrid bus arbitration policy that prevents starvation phenomenon presented in fixed priority and effectively assigns a priority to each master by mixing fixed priority and round-robin arbitration policies. The proposed arbitration policy and the others were implemented through Verilog and mapped the design into Hynix 0.18um technology and compared about gate count and design overhead. In the results of performance analysis, we confirm that our proposed policy outperforms the others in the aspect of design complexity, timing margin, bus utilization, starvation prevention, request cycle and so on.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.2B
• /
• pp.137-145
• /
• 2006

In this paper, we propose a multicast P2Patching technique in the application layer. The P2Patching technique serves VOD stream effectively with Patching in P2P environment. The P2Patching provides multicast tree construction technique that reduces the server load and minimizes the start delay with extended multicast technique. And we provide a fast recovery technique by tree failure and dynamic buffering scheme that guarantees the continuous streaming by frequent tree disconnections. Comparing the method with that of $P2Cast^{[12]}$, we obtained the better performance by our scheme in terms of average join count, service rejection probability, tree recovery failure and buffer starvation. The average about 16$\%$ of the improvement is shown by the simulation. Most of all, the performance of buffer starvation and average join count shows a significant improvement than that of P2Cast.

• The Transactions of the Korea Information Processing Society
• /
• v.5 no.12
• /
• pp.3077-3087
• /
• 1998

본 논문에서는 분산 멀티미디어 환경에서 데이터 크기의 가변성과 네트워크의 부하변동으로 인하여 발생하는 미디어 불연속성을 방지하면서 동기화를 보장하는 클라이언트 버퍼 안정화 알고리즘을 제안한다. 제안 알고리즘은 클라이언트의 상황에 맞도록 산출된 미디어스크림의 제어량과 제어시간이 포함된 제어 메시지를 멀티미디어 서버로 전송함으로써 클라이언트 버퍼를 안정화 시킬 수 있다. 클라이언트는 주기단위로 수신측의 자원과 네트워크의 상태를 예측하여 오버런 혹은 스타베이션이 발생할 징후가 보이면 제어 메시지를 생성하는데, 제어값들은 클라이언트의 상황과 멀티미디어 서버의 미디어스트림전송 메커니즘 및 네트워크의 상태를 고려하여 생성된다. 본 알고리즘의 특징은 미디어스트림의 과도한 드롭으로 인한 질 저하를 방지하며, 특히 네트워크의 부하변동이 클 경우에는 빠른 속도로 시스템을 안정화시킨다.