Adaptive MIMO Radar Architecture for Energy-Efficient Wireless Sensing in the D-Band

The D-band offering an untapped wide bandwidth is promising for high data rate communication and high-resolution wireless sensing. However, these potentials are hindered by the low performance and energy efficiency of the D-band circuits and systems. We present an adaptive multi-input multi-output (MIMO) radar architecture for energy-efficient wireless sensing in the D-band, leveraging a reconfigurable 2D array of radar transceiver front-ends, a scaling approach for the receiver (RX) signal-to-noise ratio (SNR) and the transmitter (TX) output power ($P_{\rm TX}$) with target distance, and dynamic selection of the direction-of-arrival (DOA) estimation algorithm. The reconfigurable radar array, providing an adaptive radar resolution, enhances the energy efficiency by reducing power consumption in the radar RF front-end and lowering the computational complexity in the radar back-end. The RX SNR and the TX output power are scaled with the distance as ${\rm SNR} \propto d^{-p}$ and $P_{\rm TX} \propto d^{4-p}$, where $0 < p < 4$, leading to more efficient resource allocation in varying target distance conditions. Additionally, DOA estimation results using MUSIC and MVDR algorithms indicate that the optimum algorithm, in terms of the accuracy and computational complexity, should be selected based on the number of radar array elements. Furthermore, we develop a hardware model for the MIMO radar RF front-end to evaluate the power consumption of the TX, RX, and local oscillator (LO) distribution network. It is shown that the power consumption of the LO distribution network, which can dominate the power consumption for a large MIMO radar, can be minimized through a distribution strategy for LO amplifiers employed for compensating passive losses. Performance of the adaptive MIMO radar is evaluated in the free-space and the through-wall indoor sensing scenarios in the D-band.

💡 Research Summary

**
The paper addresses the critical challenge of achieving energy‑efficient high‑resolution sensing in the D‑band (110–170 GHz), where the wide available bandwidth promises multi‑hundred‑Gbps communication and millimeter‑scale radar resolution, but existing silicon‑based implementations suffer from low transmit power, poor efficiency, and high noise figures. To overcome these limitations, the authors propose a comprehensive adaptive MIMO radar architecture that jointly optimizes hardware configuration, transmit power scaling, and direction‑of‑arrival (DOA) estimation algorithms.

Key contributions are as follows:

1. Reconfigurable 2‑D Virtual Array – A uniform 2‑D array of N_TX × N_RX elements (virtual array size N_TX·N_RX) is realized on a compact chip (λ/2≈1 mm spacing allows a 100‑element array within a 1 cm × 1 cm footprint). By disabling selected TX or RX front‑ends, the effective array size can be reduced on‑the‑fly, directly lowering RF power consumption and the computational load of the digital back‑end (which scales roughly with the number of active elements).

2. Distance‑Dependent SNR and Transmit Power Scaling – Starting from the radar range equation, the authors derive a generalized scaling law:
   \