Multi-Gigabit communication: the ADC bottleneck<sup>1</sup>

Abstract

The economies of scale in modern communication systems are enabled by architectures that take advantage of Moore's law to implement most transceiver functionalities in digital signal processing (DSP). The bottleneck in scaling such ldquomostly digitalrdquo architectures to multi-Gigabit rates becomes the analog-to-digital converter (ADC): high-speed, high-precision ADCs are either not available, or are too costly and power-hungry. In this paper, we report on recent results on two approaches towards addressing this bottleneck. The first is to simply use drastically low-precision (1-4 bit) ADCs than current practice. This could be suitable for applications that require limited dynamic range (e.g., line-of-sight communication using small constellations), but there are fundamental and algorithmic questions as to whether all the functions of a communication receiver can be realized with such a significant nonlinearity early in the processing. The second is to use a time-interleaved ADC, where a large number of low-speed, high-precision ADCs are employed in parallel to realize a high-speed, high-precision ADC. This is more generally applicable to applications requiring large dynamic range (e.g., large constellations and/or dispersive channels), but the important question is how to effectively address the mismatch between the component ADCs, which leads to a performance floor if left uncompensated.