Serial acquisition of PN sequences in direct sequence spread spectrum communication systems

We consider serial search schemes for the acquisition of a pseudo-noise (PN) sequence in a direct-sequence spread-spectrum (DS/SS) communication system. Serial search acquisition schemes are by far the most commonly used class of acquisition schemes, due to the ease with which they are implemented. The main assumptions invoked in this thesis are that no data modulation is present during the acquisition process, and that the RF carrier has already been perfectly acquired, both in frequency and phase. Within this context, two basic models are considered, the chip synchronous model and the chip asynchronous model.In the chip synchronous model it is assumed that the chip boundaries of the received PN sequence are known to the receiver. Under this model, the unknown timing offset between the received PN sequence and the local PN sequence is an integer multiple of the chip duration. We extend the results of previous work for the chip synchronous model, in which the PN sequence arising under the out-of-phase hypothesis is modeled either as a random sequence or a zero sequence. In particular, we show that the stopping times for the sequential probability ratio tests (SPRTs) used in the testing stage are exponentially bounded random variables when actual periodic spreading sequences are employed. Moreover, we study the mean acquisition time of the overall serial search scheme, and show that the scheme based on the random sequence model offers considerable savings in mean acquisition time for moderate to large values of signal to noise ratio (SNR).A more general model is the chip asynchronous model, in which the chip boundaries of the received PN sequence are not known to the receiver. Thus, measured in units of the chip duration, the unknown timing offset between the received PN sequence and the local PN sequence is modeled as a real number. We propose and study a novel serial search acquisition scheme for this model, in which the integer part and the fractional part of the offset are estimated separately. This is accomplished by splitting the available observations into two sets. One of these sets of observations is used to estimate the integer offset via serial search, and we show that the other set may be used to estimate the fractional offset simultaneously. We consider the design and performance of sequential probability ratio tests (SPRTs) for use in the testing stage which are based on both the random sequence and zero sequence models. In particular, we have obtained an SPRT based on the random sequence model which yields near optimal performance regardless of the (unknown) fractional offset. We prove that the stopping times of the SPRTs are exponentially bounded for any value of the fractional offset, provided the spreading sequence is periodic. A modified verification stage is presented such that the verification error probability satisfies any given specification. Moreover, as in the chip synchronous case, we show that the acquisition scheme based on the random sequence model offers considerable savings in mean acquisition time for moderate to large SNR. (Abstract shortened by UMI.)U of I OnlyETDs are only available to UIUC Users without author permissio