This paper analyzes the feasibility of deep convolutional neural networks (DCNN) for accurate ultra-wideband (UWB) angle of arrival estimation that is robust against hardware imperfections. To this end, a uniform linear array with four antenna elements is leveraged and a DCNN approach is proposed and compared with traditional approaches, such as MUSIC and phase difference of arrival estimators, for different environments, number of available channel impulse responses, and polarization mismatches, in terms of absolute value of error and computational complexity. The proposed approach outperforms the traditional approaches up to 80 • error reduction at a computational complexity increase of only 10% compared to MUSIC.

No abstract

A low-power IEEE 802.15.4z high-rate PHY (HRP) compatible coherent transmitter is described. The proposed transmitter uses a digital polar architecture with fixed amplitude steps in the power amplifier and asynchronous time-discrete pulse shaping. The pulse-shaping unit consists of a finite-impulse response (FIR) filter using current-starved inverter-based delay taps that can be calibrated on-chip. An injection-locked ring oscillator (ILRO)-based frequency synthesis enables wideband operation from 3-to 10-GHz frequency bands. The ILRO also allows for duty-cycled coherent mode operation with 2-4-ns phase locking time and binary phase modulation is applied directly on the oscillator. The on-chip digital front end enables duty cycling (DC) of analog front-end modules with a granularity of 2 ns. Implemented in 28-nm CMOS process, this chip is measured to consume 4.9-mW power in nominal mode with IEEE 802.15.4z high pulse repetition frequency (HPRF) compatible data rate of 6.81 Mb/s compliant with major spectrum mask regulations for channels 5 and 9. With DC of the oscillator enabled in the energy-efficient mode, a power consumption of 430 µW is achieved for packets compatible with legacy pulseposition-modulated IEEE 802.15.4a standard with a data rate of 27.2 Mb/s.