http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Timestamp를 이용한 AODV 라우팅 알고리즘의 성능 개선
이재석,박혜련,이기현,석정봉,Lee Jae-Suk,Park Hae-Ryeon,Lee Kee-Hyun,Suk Jung-Bong 한국통신학회 2006 韓國通信學會論文誌 Vol.31 No.8A
RFC3561로 공식 승인된 현재의 AODV 라우팅 방식은 노드 간 패킷 전달시간으로 40ms의 고정 값으로 설정되어 있다. 이 값과 TTL 값을 이용하여 RREQ 메시지가 목적 노드에 도착 한 후, RREP 메시지가 소스 노드에 되돌아오는 시간을 계산하게 되는데, 여기서 고정된 값을 사용하기 때문에 망의 효율이 저하되는 문제가 발생한다. 본 논문에서는 이러한 문제점을 개선하기 위하여 현재의 AODV 방식에서 RREQ 메시지에 timestamp 필드를 설정하여 패킷 발생 시간을 기록하여 전송하고, 이것을 받은 노드는 현재 시간과 패킷 내 timestamp 값을 이용하여 노드간의 패킷 전송시간을 망의 트래픽 상황에 따라 동적으로 계산하는 방식을 제안한다. 제안 방식의 정당성을 입증하기 위해, 제안 방식을 현재 방식과 함께 컴퓨터 모의실험을 수행하였으며, 그 결과 제안방식이 현재 방식보다, 특히 노드 이동성과 트래픽 부하가 클 경우, 우수한 성능을 나타냄을 증명하였다. One of the seminal operation parameters of AODV is the node-to-node travel time of data, which is currently set to fixed value of 40 ms in the RFC3561 and widely used to compute other parameters such as the expected round-trip time of the connection setup message, etc. We have naturally thought the network performance could be improved by dynamically varying the node-to-node travel time with respect to the traffic condition in the networt rather than using the fixed value, which motivates this work. To this end, we apply the idea of using timestamp; every node places the current time in the message before sending it out, and the receiver node computes the node travel time based on the moving average algorithm by considering not only the current value but also the previous ones in an accumulated and exponentially decreasing fashion with time. We evaluate the performance of the proposed scheme in respect of the number of RREQ messages generated, throughput, and delay as a function of traffic load and node mobility, and compare the result with the original AODV scheme. The results show that the proposed scheme presents noticeable performance improvements, expecially under the condition of high node mobility and high traffic load.
李演洙,金泰勳,朴惠蓮,李基鉉 明知大學校 産業技術硏究所 2003 産業技術硏究所論文集 Vol.22 No.-
In this paper, we deal with the Negative Acknowledgement(NACK)-Implosion problem that arises in the multicasting environment, and present a scheme to cope with this problem, the modified version of the Scalable Reliable Multicast(SRM) protocol, called SRM-K. This motivates this work in which the NACK message are generated depending on the congestion level inside the network. The occurrence of congestion is determined by estimating the round trip time (RTT) for each message. As an indication of the congestion level, we introduce a counter in each node, whose value K in dynamically adapted to the change of the network situation, namely increases when the congestion builds up and decreases otherwise. The proposed SRM-K scheme is evaluated by performing the simulation, showing that it outperforms the SRM, in particular, in the scenario when the network situation dynamically varies.
金太奐,金太勳,朴惠蓮,李基鉉 明知大學校 産業技術硏究所 2003 産業技術硏究所論文集 Vol.22 No.-
Abstract - Recently, the IP multicast providing point-to-multpoint delivery has received wide concerns due to the ever increasing applications based on it, such as video on demand, IP broadcasting service, and so on. However, most of the exiting multicast protocols have a scalability problem in which the network performance degrades as the increase of the network size. Among the existing protocols, the Protocol Independent Multicast(PIM) protocol has been known as one the effective protocols that handles the scalability problem. Especially, the PIM-SM is more suitable for the realistic situation of the current Internet. The purpose of this paper is to examine the effect of this threshold value on the network performance in respect of the packet loss for a variety of network topologies.