We analyze a new interference suppression structure in direct-sequence spread-spectrum (DSSS) systems that is based on two dimensional (frequency- and time-domain) notch filtering. In the first step, the filter is realized in the form of variable-step (function of estimation error-power /spl nu/) least mean square algorithm. This modification will improve the system performance in the transient period by making this period shorter. Unfortunately, in the steady-state regime, the performance will be slightly degraded. This suggests a modification that will include an algorithm with /spl nu/=2 in the transient state and /spl nu/=1 in the steady-state operation. Further modification of the algorithm will use time notching of the code division multiple-access (CDMA) receiver in the period when estimation error is larger than a given threshold. A comprehensive analysis and comparison of these modifications are presented and discussed. These results can be used either for performance enhancement of DSSS systems in the presence of jamming or for CDMA overlay-type mobile communication networks. In the latter case, the interfering signal comes from another user employing narrow-band modulation. The analysis does not include fading so that the results can be used in the case when the narrow-band modulated signal is not significantly distorted by fading. An example is line-of-sight microwave communications, which happens to be one of the most feasible applications.

This paper explains the realization and the performance of a digital stepping correlator in a wideband radio channel measurement system. Radio channel sounding is based on the estimation of the radio channel by use of maximum length sequence (m-sequence) as an effective sounding signal. Signal is detected using the stepping correlator receiver. The received signal is not converted to the baseband but the sampling is done direct from the IF-frequency. Digital implementation gives also a possibility to use a windowing property when those parts of the impulse responses that do not contain signal power are abandoned during measurements. The windowing enables high radio channel sampling rates.

This paper introduces analysis and simulation results of a DS spread spectrum system in which data detection is based on a matched filter. An approximation is made of the probability of bit error in the AWGN channel and with continuous tone jamming when a complementary code pair is used as the spreading code. It is shown that the median filter causes a loss of about 3 dB in the AWGN channel, but the dramatic effect of continuous jamming can be efficiently eliminated using a median filter.

The linear equivalence of the binary Golay complementary sequence pairs is studied from the viewpoint of feedback shift-register generators. Sequences that are synthesized by the concatenation and interleaving methods from the members of a basic code pair (the kernel) of lengths 2, 10 and 26 are considered. It is demonstrated that by the use of the Berlekamp-Massey shift-register synthesis algorithm the linear complexity value of complementary sequences is at least 3/4 of the sequence length. For some sequence pairs the linear complexity value can be even 0.98 times the sequence length. In the light of these results complementary sequences are considered suitable for information security applications employing the spread-spectrum (SS) technique in which strong non-linearity and good statistical properties are required for pseudonoise (PN) sequences.

This paper introduces the analysis and simulation results of a DS spread spectrum system in which the principle of matched filter synchronization for acquisition of code synchronization is used. Upper bounds (due to the upper bound of P/sub d/) of mean acquisition time of coherent matched filter synchronization in continuous tone jamming, when complementary codes are used as the spreading codes, are indicated. The idea of using median type filtering as an aiding device for the matched filter is presented. It is shown that continuous jamming can have a very dramatic influence on mean acquisition time if the imbalances of the codes have large values, but the influence is totally eliminated by using a median filter.>