In order to increase the link capacity in the wideband direct sequence code division multiple access (W-CDMA) reverse link, employing a parallel-type coherent multistage interference canceller (COMSIC) is more practical than using a serial (successive)-type due to its inherent advantage of a short processing delay, although its interference suppression effect is inferior to the serial-type. Therefore, this paper proposes a parallel-type COMSIC with iterative channel estimation (ICE) using both pilot and decision-feedback data symbols at each canceling stage. Computer simulation results show that the required average transmit E/sub b//N/sub 0/ for obtaining the average BER of 10/sup -3/ using the three-stage parallel-type COMSIC with ICE using pilot and decision-feedback data symbols after forward error correction (FEC) decoding (number of iterations was 2 at each stage) was decreased by approximately 2.1 dB compared to a COMSIC with channel estimation using only pilot symbols, with the spreading factor (SF) of 16 and when the number of active users was 20 (thereby the capacity normalized by the processing gain (Pg) becomes 42%). The results also show that the required average transmit E/sub b//N/sub 0/, loss of the parallel-type COMSIC using ICE at the average BER of 10/sup -3/ compared to the serial-type was negligible (i.e., within 0.2 dB) due to the significantly improved channel estimation and decreased data decision error, while achieving a much shorter processing delay irrespective of the number of active users.

The combined effect of coherent RAKE combining using the weighted multislot averaging (WMSA) channel estimation filter and closed-loop fast transmit power control (TPC) in the 4.096 Mchip/s direct sequence code division multiple access (DS-CDMA) mobile radio reverse link is experimentally evaluated. The WMSA channel estimation filter utilizes periodically transmitted pilot symbols (four pilot symbols are time-multiplexed in each 40-symbol time slot). Its observation period is extended to 2-K slots in order to improve the accuracy of the channel estimation. The fast TPC is based on the measurement of signal-to-interference plus background noise ratio (SIR) using pilot symbols. Laboratory experiments show that the use of the K=2 WMSA channel estimation filter reduces the required E/sub b//I/sub 0/ at the average BER of 10/sup -3/ by approximately 0.5 dB compared to use of the linear interpolation filter, and that the required E/sub b//I/sub 0/ is minimized when the SIR measurement interval is M=10 symbols (one slot TPC delay). It was also clarified that SIR-based TPC works satisfactorily when two users with different information data rates, i.e., SF, independently employ fast TPC. Field experimental results obtained in an area nearby Tokyo showed that the average BER of 10/sup -3/ is achieved at the target E/sub b//I/sub 0/ per antenna of approximately 2.5 dB by using four-finger branch RAKE and two-branch antenna diversity. Although the target E/sub b//I/sub 0/ to achieve same BER, when there is one interfering user with a fourfold greater transmit power than that of the desired user that independently employs fast TPC, is almost the same as that in the single-user case, the mobile transmit power is increased by 1.0-2.0 dB due to the increased MAI. These results indicate that the combination of coherent RAKE combining and fast TPC works well in practical multipath fading channels.