To reduce the TCP flow processing cost, some bit patterns selected from the TCP/IP packet can be used as TCP flow identification. Based on the entropy and randomness analysis of the distribution of sequence number （S...To reduce the TCP flow processing cost, some bit patterns selected from the TCP/IP packet can be used as TCP flow identification. Based on the entropy and randomness analysis of the distribution of sequence number （SN） and acknowledgement number （AN） in the first packet of a TCP flow, this paper proposes a new uniform TCP flow identification by sequence and acknowledgement number （FIDSAN） to the heavy-tailed IP or TCP traffic. The experimental results suggest that some bits in the TCP sequence number and acknowledgment number can be selected out as flow ID with acceptable confliction probability. The bit length of flow ID selected under given confliction probability can be conducted from an equation deduced from the observing window and flow ID range. FIDSAN has low computation cost in the comparison with the traditional methods, such as 5-tuple, CRC, and Checksum etc.展开更多
The conservative Additive Increase Multiplicative Decrease mechanism of traditional TCP causes the link under-utilization in the Wide Area Networks(WANs) due to the WANs' intrinsic nature of high latency and high ...The conservative Additive Increase Multiplicative Decrease mechanism of traditional TCP causes the link under-utilization in the Wide Area Networks(WANs) due to the WANs' intrinsic nature of high latency and high packet loss.To alleviate the problem,we present the design and implantation of STAG,an Acceleration Gateway with Split-TCP in the paper.STAG is built on embedded network equipment and acts as a transparent proxy.In STAG,a new improved congestion control method named Rapid TCP is adopted,which determines whether or not to decrease the congestion window based on the packet loss trend.In particular,in the fast recovery phase,it chooses different window adjustment strategies based on the current size of congestion window to achieve higher utilization.The performance validation of STAG is done on both our emulation testbed and the real wide area network.The results show that STAG with Rapid TCP effectively adapts to the high loss network environment and significantly speeds up the applications without loss of fairness.展开更多
We propose an adaptive fractional window increasing algorithm (AFW) to improve the performance of the fractional window increment (FeW) in (Nahm et al., 2005). AFW fully utilizes the bandwidth when the network is idle...We propose an adaptive fractional window increasing algorithm (AFW) to improve the performance of the fractional window increment (FeW) in (Nahm et al., 2005). AFW fully utilizes the bandwidth when the network is idle, and limits the op-erating window when the network is congested. We evaluate AFW and compare the total throughput of AFW with that of FeW in different scenarios over chain, grid, random topologies and with hybrid traffics. Extensive simulation through ns2 shows that AFW obtains 5% higher throughput than FeW, whose throughput is significantly higher than that of TCP-Newreno, with limited modi-fications.展开更多
文摘To reduce the TCP flow processing cost, some bit patterns selected from the TCP/IP packet can be used as TCP flow identification. Based on the entropy and randomness analysis of the distribution of sequence number （SN） and acknowledgement number （AN） in the first packet of a TCP flow, this paper proposes a new uniform TCP flow identification by sequence and acknowledgement number （FIDSAN） to the heavy-tailed IP or TCP traffic. The experimental results suggest that some bits in the TCP sequence number and acknowledgment number can be selected out as flow ID with acceptable confliction probability. The bit length of flow ID selected under given confliction probability can be conducted from an equation deduced from the observing window and flow ID range. FIDSAN has low computation cost in the comparison with the traditional methods, such as 5-tuple, CRC, and Checksum etc.
基金supported by the National Natural Science Foundation of China(Grant nos.61173169,61103204,and 61402542)the open funding of Science and Technology on Information Transmission and Dissemination in Communication Networks Laboratory(Grant no.ITDU14010/KX142600017)the Postgraduate Innovative Research Project of Hunan Province(No.CX2012B065)
文摘The conservative Additive Increase Multiplicative Decrease mechanism of traditional TCP causes the link under-utilization in the Wide Area Networks(WANs) due to the WANs' intrinsic nature of high latency and high packet loss.To alleviate the problem,we present the design and implantation of STAG,an Acceleration Gateway with Split-TCP in the paper.STAG is built on embedded network equipment and acts as a transparent proxy.In STAG,a new improved congestion control method named Rapid TCP is adopted,which determines whether or not to decrease the congestion window based on the packet loss trend.In particular,in the fast recovery phase,it chooses different window adjustment strategies based on the current size of congestion window to achieve higher utilization.The performance validation of STAG is done on both our emulation testbed and the real wide area network.The results show that STAG with Rapid TCP effectively adapts to the high loss network environment and significantly speeds up the applications without loss of fairness.
基金Project supported by the National Natural Science Foundation of China (Nos. 60625103, 60702046 and 60832005)the Doctoral Fund of MOE of China (No. 20070248095)+3 种基金the China International Science and Technology Cooperation Program (No. 2008DFA11630)the Shanghai Science and Technology PUJIANG Talents Project (No. 08PJ14067)Innovation Key Project (No. 08511500400)the Qualcomm Research Grant
文摘We propose an adaptive fractional window increasing algorithm (AFW) to improve the performance of the fractional window increment (FeW) in (Nahm et al., 2005). AFW fully utilizes the bandwidth when the network is idle, and limits the op-erating window when the network is congested. We evaluate AFW and compare the total throughput of AFW with that of FeW in different scenarios over chain, grid, random topologies and with hybrid traffics. Extensive simulation through ns2 shows that AFW obtains 5% higher throughput than FeW, whose throughput is significantly higher than that of TCP-Newreno, with limited modi-fications.
