3G FDD W-CDMA initial synchronization follows a 3-step procedure to find a primary scrambling code while locating the frame boundary. In step 2, simplified coherent combining of secondary codes has been introduced for group and slot offset identification. The proposed method shows significant performance improvement against frequency offset. The simulation results show that there is approximately 3 dB performance improvement compared to conventional coherent integration for 6 kHz carrier offset while the other results follow almost the same values as those of conventional coherent integration.

The author proposes that one can overlay an existing GSM cellular network with a broadband CDMA cellular system utilizing direct-sequence spread spectrum alone with agile notch filters at the base station. Notch filter depths of 35 dB on receive and 20 dB in transmit were found to be adequate. Capacity calculations show that the B-CDMA overlay can approximately quadruple the nominal capacity of GSM, which is 16 channels per sector, by providing an additional 51 voice channels at 16 kbps each. Also, the author has calculated that the B-CDMA mobile transmits on average with 16 dB less power than the GSM mobile, while the CDMA base transmit power per channel would be about 5.5 dB less than the GSM base due to the lower CIR required by the B-CDMA overlay system.>

Inherent SNR limitations of CCD pseudonoise matched filters (PNMF) due to fat-zero noise, input circuit noise, thermal dark current, reset noise, and charge-transfer inefficiency are investigted. It is shown that for practicable integrated circuits using buried-channel CCD's, these device limitations will rarely degrade performance enough to compete with the code correlation self-noise.

The application of charge-coupled device (CCD) technology to spread-spectrum systems is presented. In particular, CCD pseudonoise matched filters and recursive integrators are described functionally and as semiconductor devices. Then, applications to various modulation formats are considered in terms of integrated circuit architecture. Acquisition and synchronization for directsequence and hybrid frequency-hopping spread-spectrum systems are also covered. Finally, random time-division multiple-access acquisition and data demodulation are considered.

Three network search schemes-pure flood search, twoarea search, and link-by-link search-are presented and analyzed. The average search duration and the average search traffic are the performance criteria. The flood search scheme is found to be clearly inferior. Two methods for comparing the remaining two schemes are then proposed and applied. It is found that while one method might favor the two-area scheme, depending on value judgments, the other method rates the two schemes equally.