HTML5 Webook

111/182

the packet error rate is 0%. Figure 10 shows the time variation of TCP throughput when the packet error rate is 0.002%. In Figure 9, the time to reach the maximum bandwidth by hybla is shorter than that by reno. It becomes shorter when the window size is larger. In Figure 10, in the case of nite packet error, the bandwidth to time variation is lower by reno, but it increases in hybla with some uc-tuation.In the case of zero packet error, both hybla and reno achieved the bandwidth of 250–290 Mbit/s for the window sizes of 64 MB and 128 MB. In the case of packet error of 0.002%, although the bandwidth by hybla achieved 200–260 Mbit/s, that by reno remained at low throughput of 6-36 Mbit/s. Compared with reno, hybla has high per-formance even under very delayed conditions, by increasing the window size quickly when packet errors occur.(3)TDMA transmission experiment between three earth terminalsIn order to establish the conguration for TDMA transmission experiments between three earth terminals, SDR-VSAT2 was developed. e conguration of transmis-sion experiment between the three earth terminals is shown in Fig. 11. e experiment was performed under appropri-ate satellite line conditions (no bit errors). PCs were con-nected to LAN ports of modems at each earth terminal and an iperf experiment was performed between the PCs. e TDMA slot allocation for the TDMA transmission experi-ment between the three earth terminals is shown in Fig. 12. For one PC, unidirectional transmission was performed with 3 slot columns at 100 Mbit/s and bidirectional trans-mission was performed with 6 slot columns at 200 Mbit/s.e test result of iperf to UDP was shown in Table 7. e packet error of 0% was conrmed for unidirectional transmission at 100 Mbit/s with 6 lines. It was conrmed that TDMA transmission with packet error of 0% between three earth terminals can be achieved as a mesh-type network using a 622 Mbit/s TDMA satellite communica-tion system.FiF10TCP throughput change for packet error of about 0.002%050100150200250300050100150200250300Bandwidth (Mbps)Time (sec)FiF12　TDMA slot allocation for three terminals communication01112223334445556661ＲＢWRB111222333444555666211122233344455566631112223334445556664111222333444555666511122233344455566661112223334445556667111222333444555666811122233344455566691112223334445556661011122233344455566611111222333444555666121112223334445556661311122233344455566614111222333444555666151112223334445556661:LET>SDR-VSAT1 　　　　　　2:SDR-VSAT1>LET3:SDR-VSAT1>SDR-VSAT2 4:SDR-VSAT2>SDR-VSAT1　5:SDR-VSAT2 > LET 6:LET > SDR-VSAT2FiF11Configuration of three terminals TDMA communication experimentMODEM-1MODEM-2(Slave)200Mbit/s200Mbit/sJRBReference Station(JAXA)LETSDR-VSAT-1,2JRBreceiverSatellite position,slot assign info., etc.Synchronous signalSS-TDMA(Master)WRB200Mbit/s-2PC-1-4PC-3MODEM-3(Slave)-6PC-5TabT7　Result of UDP iperf test about between three terminalsLET⇔SDR-VSAT1Eb/No100 MbpsRemarksModem-1⇒Modem-216 dB-17 dBLoss 0 %3 slots in frameModem-2⇒Modem-116 dB-17 dBLoss 0 %3 slots in frameLET⇔SDR-VSAT2Modem-1⇒Modem-316 dB-17 dBLoss 0 %3 slots in frameModem-3⇒Modem-116 dB-17 dBLoss 0 %3 slots in frameSDR-VSAT1⇔SDR-VSAT2Modem-2⇒Modem-316dB-17 dBLoss 0 %3 slots in frameModem-3⇒Modem-216dB-17 dBLoss 0 %3 slots in frame1073-4 622/1,244 Mbit/s Dual Rate High-Speed Burst Modem TDMA Satellite Communication Experiments