Code-division multiplexing

Abstract

(cont.) counterpart. Among intra-cell orthogonal schemes, we show that the most efficient broadcast signal is a linear superposition of many binary orthogonal waveforms. The information set is also binary. Each orthogonal waveform is generated by modulating a periodic stream of finite-length chip pulses with a receiver-specific signature code that is derived from a special class of binary antipodal, superimposed recursive orthogonal code sequences. With the imposition of practical pulse shapes for carrier modulation, we show that multi-carrier format using cosine functions has higher bandwidth efficiency than the single-carrier format, even in an ideal Gaussian channel model. Each pulse is shaped via a prototype baseband filter such that when the demodulated signal is detected through a baseband matched filter, the resulting output samples satisfy the Generalized Nyquist criterion. Specifically, we propose finite-length, time overlapping orthogonal pulse shapes that are g-Nyquist. They are derived from extended and modulated lapped transforms by proving the equivalence between Perfect Reconstruction and Generalized Nyquist criteria. Using binary data modulation format, we measure and analyze the accuracy of various Gaussian approximation methods for spread-spectrum modulated (SSM) signalling ...

We study forward link performance of a multi-user cellular wireless network. In our proposed cellular broadcast model, the receiver population is partitioned into smaller mutually exclusive subsets called cells. In each cell an autonomous transmitter with average transmit power constraint communicates to all receivers in its cell by broadcasting. The broadcast signal is a multiplex of independent information from many remotely located sources. Each receiver extracts its desired information from the composite signal, which consists of a distorted version of the desired signal, interference from neighboring cells and additive white Gaussian noise. Waveform distortion is caused by time and frequency selective linear time-variant channel that exists between every transmitter-receiver pair. Under such system and design constraints, and a fixed bandwidth for the entire network, we show that the most efficient resource allocation policy for each transmitter based on information theoretic measures such as channel capacity, simultaneously achievable rate regions and sum-rate is superposition coding with successive interference cancellation. The optimal policy dominates over its sub-optimal alternatives at the boundaries of the capacity region. By taking into account practical constraints such as finite constellation sets, frequency translation via carrier modulation, pulse shaping and real-time signal processing and decoding of finite-length waveforms and fairness in rate distribution, we argue that sub-optimal orthogonal policies are preferred. For intra-cell multiplexing, all orthogonal schemes based on frequency, time and code division are equivalent. For inter-cell multiplexing, non-orthogonal code-division has a larger capacity than its orthogonal