Analog Multiband: Efficient Bandwidth Scaling for mm-Wave Communication

We investigate analog multiband as a means of scaling communication bandwidths over dispersive channels: the available band is channelized into contiguous subbands in the analog domain and digitized in parallel at the receiver. The subband width is chosen such that existing analog-to-digital converter (ADC) technology provides dynamic range sufficient to capture the effects of channel dispersion and interband interference. This avoids the difficulty of scaling high-precision ADCs to large bandwidths, while allowing the use of sophisticated digital signal processing (DSP) techniques for all transceiver operations except for channelization. In this paper, we address two fundamental bottlenecks associated with this concept. The first is channelization. A direct approach using a bank of mixers with independent frequency synthesizers is power-inefficient and subject to oscillator coupling, hence we explore an alternative approach based on polyphase sampling and sampled analog fast Fourier transform (FFT), along with appropriately designed baseband filters. The second is interference due to imperfect channelization (in the interest of bandwidth efficiency, we do not use guard bands) and imperfections in analog processing. We characterize the unique structure of this interference when OFDM is used over each subband, and show that linear adaptive interference suppression on the edge subcarriers suffices to provide robust performance. MultiGigabit/s millimeter (mm) wave communication is a key application driver for this work, hence we illustrate our ideas with performance evaluation using indoor channel models developed for the IEEE 802.11ad 60 GHz standard.