Toshihiro Shikama

Mitigation of bursty packets by a TCP proxy improving TCP performance in a wired and wireless network

This paper investigates a TCP proxy that splits a TCP connection into two parts consisting of a wireless link and a wired network. The TCP proxy is effective for improving TCP performance in such a heterogeneous network including a wireless link, on which transmission errors occur. This paper describes how the TCP proxy produces large size forward data due to a packet loss on the wireless link. It also identifies a new problem that the output traffic from the TCP proxy becomes bursty due to the large size forward data. To mitigate bursts of packets by the TCP proxy this paper proposes a scheme that the proxy performs a pacing function which places a gap between two consecutive packets. Since this function is performed in cooperation with the flow control between two TCP connections, the scheme has an advantage that it requires a small amount of forwarding buffers for the pacing. Simulation results using ns-2 show that bursty packets are produced by the conventional TCP proxy and the pacing function suppresses them. The results also show that throughput is improved by the proposed scheme, while the throughput of the conventional TCP proxy suffers from packet losses in the wired network due to the bursts of packets.

Resequencing schemes for selective-repeat ARQ and their performance

SR (selective-repeat) ARQ (automatic repeat request) is employed to recover packet errors in communication environment with long propagation delay. This SR ARQ has problems of large resequencing delay and bursty packet output due to the resequencing function. In order to solve these problems this paper proposes two types of resequencing schemes and compares the TCP throughput of the schemes by simulations, where a number of upper layer flows are multiplexed over SR ARQ performed in Layer 2.

Delay analysis of the selective-repeat ARQ with the per flow resequencing

An SR (selective-repeat) ARQ (automatic repeat request) protocol is used for effective packet error recovery on a satellite channel. This SR ARQ has a problem of large delay due to the resequencing of received packets. To mitigate this problem, the PFRS (per flow resequencing) scheme was proposed, where resequencing is performed independently for each upper layer flow, while the detection of lost packets and associated retransmissions are performed based on the whole flows multiplexed over the SR ARQ. The paper models the SR ARQ protocol, where the maximum number of retransmissions is limited, by a collection of simple stop-and-wait protocols; it shows calculation results on the delay distribution of retransmission and resequencing and proves the effectiveness of the PFRS scheme.

The erasure nodes control scheme on DQDB

Erasure nodes in distributed queue dual bus (DQDB) networks are discussed. When erasure nodes are introduced into DQDB networks, the disturb user communication in light traffic. To solve this problem, it is proposed that the number of erasure nodes be changed based on the traffic conditions. The details of this method are presented. Performance characteristics obtained for the method are also presented.<<ETX>>

An architecture for connectionless data service in B-ISDN

A method for the support of a direct B-ISDN (broadband integrated services digital network) connectionless service is proposed. The method has the advantages of eliminating the load to call control function, a small routing decision delay, simple routing control, and efficient use of virtual connections. The method is compared with the virtual ring method, which is proposed by the CCITT as an example of the direct provision of connectionless service, in terms of the traffic load within the network and the processing load in the connectionless service function. It is shown that the proposed method is suitable for a large-scale network and can be advantageously used when the number of cells associated with a packet is large.<<ETX>>

SAT01-2: Service Differentiation by a Link Layer Protocol Based on SR ARQ over a Satellite Channel

This paper studies the case where multiple IP flows are aggregated over a single satellite channel and an error recovery by retransmissions is performed by Selective- Repeat (SR) ARQ. We propose MQ-PFRS (Multi-QoS Per-Flow Resequencing) ARQ that provides a differentiated service for an IP flow depending on its traffic class, such as real-time traffic (ex. UDP) or non-real-time traffic (ex. TCP). MQ-PFRS ARQ eliminates interferences among IP flows by resequencing received packets independently for each IP flow. It also controls the maximum packet delay by limiting the persistency of retransmissions and the maximum resequencing time for each packet. This paper also presents the analysis of the probability distribution of real-time packet delays. Simulation results show that the delay variation of real-time traffic is effectively controlled by proposed MQ-PFRS ARQ and the results of the analysis well capture the characteristics of the packet delay distribution. This means that MQ-PFRS is effective for handling multiple classes of IP flows and quality of real-time traffic transferred by the scheme can be predicted by the analysis.

A proposal of gateway decentralization method in wireless mesh networks and its evaluation

We proposed the session distribution method as one of the multi GW selection methods. Based on the results of the simulation, we came to the conclusion that the TCP throughput becomes larger in the session distribution method compares with the session distribution method. As the next step, we are going to make the implementation of the session distribution method.

A Multilink Protocol with the Per Flow Resequencing and Its Performance Evaluation

Procedures that provide a virtual link service aggregating bandwidths of multiple physical links are called a multilink protocol. The multilink protocol enables to utilize bandwidths of multiple links flexibly and to improve reliability of communications. This paper focuses on the resequencing function needed to preserve packet sequence integrity. The resequencing is generally performed by both each link layer protocol as well as the multilink protocol. This paper shows that a problem of the head of line (HOL) blocking occurs due to the resequencing if packets of multiple IP flows are transferred over the multilink connection. To solve this problem, this paper proposes the Per- Flow ReSequencing (PFRS) scheme for the multilink protocol and evaluates its effectiveness through analysis and simulations.

Changeover without Slot Loss of the Erasure Nodes on DQDB

This paper concerns the erasure nodes to release slots in DQDB. If the erasure nodes are introduced in the DQDB network, the capability to transfer slots is improved. However, the transit delay from the source node to the destination node is increased by the storage delay in the erasure nodes. We proposed that the number and the location of the erasure nodes are changed according to traffic condition in our pervious work. However, these changes should not disturb the valid slot streams. This paper proposes the scheme to provide this function. This paper also shows the configuration of the erasure nodes having this function. Moreover, it describes the operation algorithm to provide the changeover of a node state, where activation as well as inactivation of slot erasing function in a node is performed without loss of slots. Finally, it estimates the time of this algorithm execution.