Software Tools: MATLAB
- E. Raei, M. Alaee-Kerahroodi and M. R. B. Shankar, “Spatial- and Range- ISLR Trade-Off in MIMO Radar Via Waveform Correlation Optimization,” in IEEE Transactions on Signal Processing, vol. 69, pp. 3283-3298, 2021, doi: 10.1109/TSP.2021.3082460.).
- E. Raei, M. Alaee-Kerahroodi and B. M. R. Shankar, “Waveform Design for Beampattern Shaping in 4D-imaging MIMO Radar Systems,” 2021 21st International Radar Symposium (IRS), Berlin, Germany, 2021, pp. 1-10, doi: 10.23919/IRS51887.2021.9466196.
- N. K. Sichani et al., “Waveform Selection for FMCW and PMCW 4D-Imaging Automotive Radar Sensors,” 2023 IEEE Radar Conference (RadarConf23), San Antonio, TX, USA, 2023, pp. 1-6, doi: 10.1109/RadarConf2351548.2023.10149733.
The technology of Massive MIMO, through the deployment of numerous antennas at both the transmit (Tx) and receive (Rx) chains, has revolutionized wireless communications by significantly improving their capacity and reliability, making it a favoured option for various applications, including 5G. The technology has opened up exciting new possibilities for 4D (three-dimensional space + Doppler) imaging using mmWave radar sensors, with exceptional spatial and temporal resolution. Spatial resolution refers to the system’s ability to differentiate between objects that are in angular proximity, while temporal resolution relates to its ability to precisely capture and track changes over time. By utilizing a large number of antennas to transmit and receive signals, Massive MIMO radar can focus energy towards the intended target, minimizing interference from other sources. This enables the sensor to capture fine details about the object’s position, velocity, and direction of movement, which can be invaluable for applications such as medical imaging, autonomous driving, in-car monitoring, industrial automation, and security surveillance. As a result, Massive MIMO radar can deliver exceptional precision and accuracy, providing unprecedented levels of insight into the world around us.
Virtual channels = 8
This figure shows when a mmWave radar sensor at 60GHz operating frequency with 8 virtual channels trying to recognize two objects which are in the form of logos of SnT and IEE.
Virtual channels = 500
The above example is repeated in this figure with a similar operating frequency and bandwidth but increasing the virtual channels to 500. In the new figures, the logos (reference targets) are clearly observable. The figure can be further enhanced by increasing number of virtual channels in both directions of azimuth and elevation.
Utilizing a large number of antennas in 4D imaging radar systems raises important research questions, such as:
- Antenna Design: How to place Tx-Rx antennas to provide cost-efficient comprehensive view of the monitoring surface? How to minimize crosstalk between the Tx and Rx antenna elements?
- Waveform Design: How to generate distinct orthogonal waveforms for the large number of Tx channels? How to distinguish all transmit sources in each receive channel?
- Receiver Design: How to choose ADC sampling rate to maintain dynamic range and signal-to-noise (SNR) quality while reducing the production cost? How to process and analyse massive amounts of data generated by the system in real-time?
- System Design: How to optimize the system configuration and parameters for a specific application? How to make the system more cost-effective and scalable for large-scale collaborative deployment in various applications?
MIMO technology in radar refers to a diversity that is achieved by using different transmit waveforms, offering more degrees of freedom for probing signals compared to conventional phased arrays, where only a phase-shifted version of a waveform can be transmitted. MIMO is particularly useful when there is a separation between the transmit and receive antenna arrays, allowing for a sparse configuration. In such cases, orthogonal waveforms can be transmitted to continuously illuminate the entire scene without beam scanning during transmission, while beamforming will be performed during reception. By properly designing the sparsity between the transmit and receive arrays, grating lobes can be eliminated in two-way antenna beampattern, and maximum virtual array length can be achieved. A virtual array can result in fewer Tx and Rx chains, but still offer numerous channels, thereby significantly reducing costs, particularly in mass production. This is why MIMO is now a standard for mmWave sensors used in industrial and automotive applications.
Massive MIMO technology has been widely used in the telecommunications industry to improve wireless network coverage and capacity. One of the key benefits of Massive MIMO for communications is that it can improve spectral efficiency, which means that more data can be transmitted over a given bandwidth. This can help to address the growing demand for high-speed wireless data services, especially in densely populated urban areas where multiple users are competing for limited bandwidth. Massive MIMO is expected to play a key role in enabling the development of future wireless networks in 5G New Radio and beyond. However, its application in radar signal processing is a relatively new area of research. The integration of Massive MIMO technology with radar signal processing has the potential to revolutionize the field of target imaging and tracking, leading to more efficient and effective radar systems.
Massive MIMO radar refers to the use of a large number of antennas at both the transmitter and receiver to further improve the performance of MIMO radar systems. By introducing massive antennas, transmit beamforming can be included in the system, which can help to further improve the signal-to-noise ratio and range resolution of the radar system. However, the use of massive antennas in MIMO radar systems can also pose some limitations, such as increased power consumption and complexity, as well as potential interference issues. Several challenges related to antenna array design, waveform selection, network configuration, and efficient sampling are required to be addressed in Massive MIMO radar systems.
Antenna Array Design
The aperture size of the antenna array is a critical parameter in determining the spatial resolution and image quality of a Massive MIMO 4D-imaging radar system. When the aperture size is wide relative to the wavelength, the phase coherency of the received signals becomes important, as the phase of the signals received by different antennas needs to be properly aligned to accurately determine the location and motion of targets.
In addition, the near field effect, which is the region close to the antenna array where the electromagnetic waves behave differently from the far-field region, can also impact the performance of a Massive MIMO 4D-imaging radar system. The near-field distance is directly proportional to the squared value of the largest dimension of the antenna and inversely proportional to the wavelength. Therefore, when the aperture size is wide, the near-field region can cause distortions and artifacts in the radar image, which can affect the accuracy of target detection and tracking.
To ensure optimal performance of a Massive MIMO 4D-imaging radar system with a wide aperture size, it is essential to carefully consider both the phase coherency of the received signals and the near-field effect when designing and operating the system. This may involve implementing appropriate signal processing techniques and optimizing the antenna array design to minimize the impact of these factors on the radar image.
When the transmit array is sparsely located in a MIMO radar system, transmitting a single waveform through the antenna array can create a beampattern that requires a scanning strategy to search the surveillance area. This approach can be time-consuming and inefficient, especially in dynamic scenarios where quick responses are required. To avoid this, an alternative approach is to transmit orthogonal waveforms through the transmit antennas. Orthogonal waveforms are signals that are completely independent of each other, and they have no overlap in time, frequency, or code domains.
Orthogonality can be achieved in time, frequency, or code using techniques such as time division multiplexing (TDM), frequency division multiplexing (FDM) or code division multiplexing (CDM)/binary phase modulation (BPM) . However, TDM and FDM techniques do not fully exploit the available resources of time, and frequency, and CDM/BPM, requires slow-time (outer) coding for multiplexing which creates folding in Doppler domain. Alternatively, fast-time coding is a viable solution that efficiently uses time, frequency, and Doppler domains, but requires careful waveform design strategy to minimize harmful effects of sidelobes in a 4D-imaging system. Hence, a feasible solution may be obtained by using optimization techniques that provide nearly perfect orthogonality while complying with hardware constraints. These techniques can optimize the waveform design to maximize the orthogonality between the waveforms while minimizing the interference between them. By doing so, the Massive MIMO radar system can achieve better performance with optimum usage of resources.
When the radar system uses FMCW waveforms, the technique of de-chirp can be employed to decrease the required ADC sampling rate. The de-chirp technique divides the matched filter operation in the radar system into two parts: a mixer with a variable oscillator that generates a beat frequency signal and an FFT. By using the de-chirp technique, the ADC sampling rate in FMCW radars can be considerably reduced relative to the signal bandwidth, leading to a significant decrease in cost, particularly when the number of receive channels is large. However, due to the aforementioned limitations for multiplexing techniques of FMCW, a switch to PMCW waveforms for 4D imaging Massive MIMO radars may be necessary.
In traditional PMCW radar systems, the received signals are sampled at a Nyquist rate, which is twice the maximum bandwidth of the signal. However, in Massive MIMO 4D imaging radars, the number of antennas and the complexity of the signal processing algorithms can result in extremely high sampling rates, which can be challenging to implement cost-efficiently in practice. In this case, sub-Nyquist sampling techniques can be used to reduce the sampling rate required by the system. These techniques take advantage of the fact that the signal of interest is typically sparse in some domain, such as the time-frequency domain. By exploiting this sparsity, it is possible to reconstruct the signal from a much smaller number of samples than would be required by traditional Nyquist sampling.
Alternatively, analog correlators can also be used to reduce the sampling rate required by the system. These devices can perform the correlation of the received signal with a reference signal directly in the analog domain, eliminating the need for high-speed ADCs. Combining these two can significantly reduce the complexity and cost of the system.
Detection, Image Reconstruction, and Tracking
The high spatial resolution provided by a large number of antennas in Massive MIMO systems can be exploited to improve the angular resolution of the radar system, allowing for better detection and tracking of targets in the surrounding environment. However, designing an optimum detector in such a case where lots of antennas are available may not be straight forward, especially when dealing with complex and highly variable environments. In this case, deep learning which has shown promising results in radar detection and classification tasks, can be an alternative solution. The large number of antennas in a Massive MIMO radar system can provide a rich image-like dataset that can be used for training deep learning models. Moreover, the ability of deep learning models to learn from large and diverse datasets can be particularly advantageous in Massive MIMO radar systems, where the signal environment can be highly variable and unpredictable.
If image-like information being provided, tracking and classification of the objects can be performed more efficiently, but requires novel approaches to include Doppler information with the image-like data for a better tracking and classification.