Trade-offs of performance and single chip implementation of indoor wireless multi-access receivers

Abstract

Abstract-The performance and computational complexity of five multi-access receivers are compared A methodology is then presented for making area andpower estimates of these-algorithms for both SOfWare Programmable Dsp and dedicated direct mapped architectures. With this merhodofogy and by using eqerimental data from previous designs, the feasibirity of implementatwn of the multi-access receivers can be determined multiple portable terminals. One of the challenges in design of the algorithms for such a-link is to provide enough diversity against multipath fading. For the indoor channel, the delay spreads are usually on the order of 10-100 ns. In order to support a data rate ranging from 1-10 mits/sec per user and of time, frequency and/or spatial diversity is required.We will consider two basic approaches. The frrst is based on wideband direct-sequence CDMA, and exploits frequency diversity to resolve the multipath components at the receiver and uses multiuser detection to suppress the interference from signals intended for other users. The other approach we will investigate is to employ a narrowband solution by splitting the channel bandwidth into subbands with bandwidths small compared to the channel coherence bandwidth. The flat fading due to the destructive addition of the various paths in each of the narrow band can be combated through the use of spatial diversity using multiple antennas at the transmitter and at the receiver, which can additionally be used to increase the capacity.of users, a

Truncated Online Arithmetic with Applications to Communication Systems

“…Depending on the algorithm implementation and the desired bit error rate performance, the precision requirements of the hardware processing the received signal are typically in the 8-32 bit precision range for finite precision implementations (see Fig. 2) such as those in [10], [27], [28], with more sophisticated algorithms requiring greater precision for operations such as division and matrix inversions. This implies extraneous computations in a conventional number system as the sign is obtained only at the end due to the Least Significant Digit First (LSDF) nature of computations.…”