MIMO-Radar Signal Processing ChaiN
Introduction – System overview
In this paper we provide a short overview of a modern MIMO radar system.
The block diagram of the considered system is shown in the Fig. 1.
The system contains the following blocks:
- TX processor: This block generates the waveform which is transmitted by the radar. In this document will focus on FMCW waveforms [GR1] which are commonly used in modern radars. For MIMO radar, there are two major approaches for transmission through multiple TX antennas (the number of TX antennas is Ntx):
Time division multiple access (TDMA) - Transmission of the same waveform through multiple antennas, but at the different time instances, i.e. at the beginning the 1st TX antenna transmits the waveform and all other TX antenna are switched off, then the 2nd TX antenna transmits, while the rest are switched off, etc. See Fig. 2 (a)
Frequency division multiple access (FDMA) – Multiple orthogonal waveforms are transmitted through all the antennas simultaneously. In FDMA approach the orthogonally is achieved by the frequency shift of the waveform. See Fig. 2 (b).
- RF front end: This block serves as a bridge between analog signals and digital signals. For the TX chain, this block converts digital signals to RF signals, amplifies them and send them to the antennas. For RX chain, the block receives signals from the RX antennas, amplifies and filters them, and converts them to digital samples. Commonly, for FMCW radars, the RX signals are mixed with the TX signals to extract the beat frequency, which translates directly to range
- MIMO RX processing: This block receives signals from Nrx RX channels. Using the orthogonality property of the TX waveforms, this block extracts Ntx signals from each RX channel. The total number of signals after MIMO processing is Ntx * Nrx, these signals can be seen as the output of a virtual array of size Ntx * Nrx.
- Range-Doppler processing: This block estimates target distances by performing spectral analysis of the beat frequencies. In addition, by transmitting a sequence of FMCW pulses, we can track the phase change of the reflections from each target. For moving targets this phase change can be modeled as a complex exponent which frequency is known as the Doppler frequency. These Doppler frequencies are associated with the target radial velocities and they are estimated by the Range-Doppler processor. The output of the Range-Doppler processor is a 2D map where targets are represented by energy peaks. For a MIMO radar, the Range-Doppler processing is applied for each (virtual) channel.
- Detection: This block detects peaks in the Range-Doppler map which are associated with potential targets. The output of this block is a list of potential targets with their ranges and radial velocities.
- Direction of arrival (DOA) estimation: This block estimates direction of each potential target. In order to do this, it uses signals received from the multiple channels (virtual MIMO channels). The DOA estimation is based on the fact that the phase change of the signals at the different channels is due to direction of the target. This phase change is associated with a spatial frequency of the targets which should be estimated by the DOA algorithm. More on direction of arrival estimation can be read on in this post.
- Tracker: This block receives list of potential targets with their range, velocity and DOA. Using history of the previous location (and velocity) of the targets and current measurement, the block estimates the current location and velocity of each target.
- Target Classification Processing: This is the part that uses past and present features gathered on each tracked target in order to classify the target. For example differentiating between a pedestrian and a bicycle on the road, a car or a truck, drone or bird, etc. Example for features can be motion trajectory pattern, RCS fluctuation, micro-Doppler signatures and more.
About the Author:
Ilia Yoffe is a principal research scientist in AlephZero Consulting. He is an Algorithms scientist with extensive industry experience. Holds a Ph.D. in electrical engineering. Published more than 15 papers in top journals and conferences. Specializing in digital communication (physical layer), spatial signal processing and radar signal processing. Experienced with machine learning, deep learning, computer vision. Experienced with communications standards including 802.11 (Wi-Fi and Wigig) and 5G, as well as reverse engineering communications protocol based on air-recordings on downlink/uplink.