An Enabling Technology for 6G Wireless Communications


Grant Data
Project Title
An Enabling Technology for 6G Wireless Communications
Principal Investigator
Professor Li, Can   (Co-Principal Investigator (Co-PI) (for projects led by other university))
Co-Investigator(s)
Professor Huang Kaibin   (Co-Investigator)
Professor Wang Zhongrui   (Co-Investigator)
Duration
36
Start Date
2023-06-01
Amount
877165
Conference Title
An Enabling Technology for 6G Wireless Communications
Keywords
1) THz Integrated Circuits Random-Access Memory 2) Metasurfaces 3) In-Memory Computating 4) Resistive Random Access Memory 5) 6G
Discipline
Microwave and Terahertz Engineering
Panel
Engineering (E)
HKU Project Code
C1009-22G
Grant Type
Collaborative Research Fund (CRF) - Group Research Project 2022/2023
Status
On-going
Objectives
1. This project aims to demonstrate a short-range THz communications link based on IEEE Standard 802.15.3d as an enabling technology for the 6G mobile networks. Funded by a Sixth Round Theme-based Research Scheme project, we have developed several THz radiator- source and beam-steering array ICs. To form a communications link, we need to incorporate a THz mixer and amplifier into the chip with carrier modulation via an X-band in-phase quadrature (IQ) mixer. Specifically, we propose to 1) design a THz power amplifier working at 300 GHz, 2) design a second-harmonic THz mixer, 3) design a transmitter and receiver, and 4) design 4-channel THz transmitters and receivers with 1×4 and 2×2 array chips fabricated by 65-nm CMOS technology for power enhancement, beamforming, and beam manipulation.2. We aim to employ the designed radiator ICs to enable the multiple-input and multiple-output (MIMO) and orbital angular momentum (OAM) technologies that enhance the THz wireless link's capacity. Furthermore, we will develop metasurface technologies to extend the communication range and manipulate the properties of the radiated and received waves. We will first focus on designing antenna arrays using the ICs for power combining to increase the signal-to-noise ratio and enable beam manipulation, including steering, focusing, and polarization reconfiguration. We will then demonstrate THz MIMO and OAM multiplexing. For the latter, a metasurface lens will be designed for a non-diffractive beam with the multiplexed OAM modes generated by a circular chip array. We will also design a metasurface lens to generate OAM modes and their multiplexing instead of the signaling control of the chip array. The radiator chips will be used as the source for high-gain antennas. By controlling the amplitude and phase of the chips, we can focus the energy on a spot whose location from the source can be dynamically tuned. Through a metasurface, the spot location and thus the transmission range can be further extended. The distance will be doubled when a similar arrangement is made at the receiver.3. We aim to realize ultra-fast and ultra-energy-efficient baseband processing using in-memory computing to overcome the ""von Neumann bottleneck"" in traditional designs. Such baseband processing in the analog domain facilitates integration with a THz phased array or a THz reconfigurable surface. The method will involve the design of customized crossbar arrays of RRAM devices for implementing a number of essential baseband functions, including 1) codebook-based MIMO beamforming, 2) codebook-based MIMO precoding, 3) dual-model reconfigurable linear receiver for zero-forcing (ZF) and minimum-mean-squared-error (MMSE) MIMO detection, and 4) robust orthogonal frequency division multiplexing (OFDM) to address issues in practical RRAM arrays. The design will build on RRAM crossbar arrays, capable of ultra-fast vector-matrix multiplication, and content addressable memory (CAM), which supports ultra-fast table look-up (or codebook search). New RRAM compatible signal processing techniques will be designed to address practical issues such as RRAM write- noise, defective devices, instability of matrix-inversion circuit, and easy circuit reconfiguration for receiver mode switching.4. We aim to develop high-precision RRAM devices and versatile CAM hardware to improve the performance of in-memory baseband processing, design a high-performance digital- analog interface and fabricate different baseband modules into a prototype chip. First, the high-precision (512 conductance states) of RRAM devices will mitigate the adverse effects of write-noise on baseband performance. It will be achieved by identifying noise-inducing environmental factors, engineering protection mechanisms as well as optimization of material stack. Second, versatile RRAM-based CAM will be designed to be capable of optimization using different distance metrics (i.e., cosine, L1, L2, and sub-space distance), which facilitates beamforming/precoding codebook search and other baseband optimization. Third, we will design 1) a high-performance analog-to-digital converter (ADC) using compact and power-efficient time-based architecture and 2) a high-precision digital-to-digital converter (ADC) using the delta-sigma modulation with a high oversampling ratio and noise shaping using a switch-capacitor circuit. In the final stage, we will fabricate a prototype IC by integrating RRAM arrays and CAM circuits with CMOS transistors as selectors in the former using an approach combining the use of a commercial foundry and a ""back-end of line"" process in a university cleanroom.5. We will perform system integration of the THz transmitter/receiver ICs with the in-memory baseband processing and evaluate the efficacy of the THz wireless links in the scenarios of MIMO, beamforming, and OAM multiplexing. The baseband analog output will be modulated by an X-band IQ mixer and then divided into four channels with variable amplitude and phase provided by the beamforming control. Each of these four channels will be up-converted on- chip to 300 GHz via a 2nd harmonic mixer. Then the information-carrying THz waves is radiated into the air after passing through an amplifier. A similar arrangement is made in the receiving chain without a low-noise amplifier. Specific efforts also include chip packaging, wire bonding to testing circuits on printed-circuit-boards, placement of the THz chips for beamforming, beam manipulation, OAM generation, and alignment with metasurface lenses