HTML5 Webook

176/182

e eective throughput shown in Fig. 10 (a) is dier-ent from that shown in Fig. 8 (a). e throughput of HpFP is slightly lower than the target throughput of 10 Mbps, which is a preferable result. Although some packet loss occurs as shown in Fig. 10 (b), it does not happen con-tinuously. In addition, the RTT is almost stable, except the beginning and the end of the experiment. e decrease in throughput in Fig. 10 (a) oen occurs with the occurrence of packet loss in Fig. 10 (b).Figure 10 (c) shows the time variation of the number of error detection by CRC. e detection is done every 200 ms. A few errors are detected at the beginning, but the number of error detection is almost zero aerwards. It means that the error detection by CRC is quite small as shown in Fig. 10 (c). Most of the packet loss in Fig. 10 (b) are not caused by the bit error.4.2.2Case of occurrence of network disturbancee results of measurement during the network distur-bance shown in Fig. 9 are shown in Figs. 11 and 12. e measurement shown in Fig.11 is conducted eight minutes before the measurement shown in Fig. 10 (that is 12 min-utes aer that shown in Fig. 8). e measurement shown in Fig. 12 is conducted three minutes aer the measure-ment shown in Fig. 10. e formats of both the gures are the same as that of Fig. 10. is experiment is conducted before and aer the measurement shown in Fig. 10. It means that the quality of WINDS communication lines varied during a short time.e eective throughputs shown in Figs. 11 (a) and 12 (a) are signicantly smaller than the target throughput of 10 Mbps for HpFP, which is the same as the case of Fig. 8 (a). is is because the eect of the packet loss shown in Figs. 11 (b) and 12 (b) is large. It is known that HpFP achieves high transmission performance at a certain PLR [3]. However, if the PLR is larger than 10% in geosta-tionary orbit satellite communication in which the RTT is 500 ms, the throughput decreases [8]. Moreover, in Figs. 11 (b) and 12 (b), RTT is stable at around 500 ms. is implies that the cause of the decrease in throughput in Figs. 11 (a) and 12 (b) is not overow of the buer in the modem or in the transponder.Figures 11 (c) and 12 (c) show the time variation in the number of error. e time interval of data acquisition is the same as that in Fig.10 (c) of 200 ms. e error detection FiF10　Experiment of no occurrence of network disturbanceFiF9　WINDS experiment time lines: (a) first half of the experiment and (b) second half of the experiment2016/12/27Time14:0013:0013:3013:16-13:174K video streaming (Fig. 7) preliminary experiment (Fig. 8)13:28‐13:29At the time of occurrence of network disturbance (Fig. 11)13:39-13:40At the time of no occurrence of network disturbance (Fig. 12)13:36‐13:37At the time of no occurrence of network disturbance (Fig. 10)(a)2016/12/27Time17:0017:0017:3017:20-17:21Target throughput 100MbpsCRC no check (Fig. 14)17:21-17:22Target throughput 100MbpsCRC check (Fig. 15)17:25-17:26Target throughput1400MbpsCRC check (Fig. 16)(b)(a)(b)(c)3 Ultra-High-Speed Satellite Communication Technology172　　　Journal of the National Institute of Information and Communications Technology Vol. 64 No. 2 (2017)