With the rapid development of Advanced Driver Assistance Systems, or ADAS, an increasing share of new vehicles has been mounted with mmWave radar sensors these years. The problem follows and becomes even more apparent that radar signals among different vehicles will interfere with each other.
In response to this, Calterah comes up with a solution powered by its Alps radar SoC. Before going further, let’s first have a look at the commonly seen signal interference between radar sensors. Fig. 1 shows how FMCW radar signals are typically interfered. Suppose a working radar sensor receives chirp signals from a nearby radar sensor and the interference signal frequency approximates to the frequency of the working radar sensor. Then the interference signals will go into the IF bandwidth of the working radar sensor, as shown in Fig, 2. As seen, the interference signals affect the waveform of working radar signals a lot and therefore it is very likely that desired target signals are missed or false target occur.
To address this issue, baseband of Alps SoC provides several anti-interference features, including the Frequency Hopping mode, Chirp Shifting mode, Phase Scrambling mode, and the interference mitigation function.
In this mode, start frequencies of chirps inside a frame are randomly altered by the XOR chain. Fig. 3 shows the chirp signal in Frequency Hopping mode.
When XOR chain outputs 0, the start frequency of a chirp will not be altered. When the output is 1, it is altered;
In an environment where there is interference with the same frequency, randomly altering the start frequencies of chirps can make the mixer’s output stand out of the IF bandwidth and then be filtered out.
In this way, the energy of received interference signals in a frame will be halved, provided start frequencies of half of the chirps in a frame are altered.
For the interference signals that go into the baseband, their energy will be distributed in the spectrum of the entire 2D-FFT since they randomly occur among chirps in a frame, and as a result no false target will occur due to clustering.
In Chirp Shifting mode, start points of ramp-up time of chirps inside a frame are randomly altered by the XOR chain.
When XOR chain outputs 0, the start point of ramp-up time of a chirp will not be altered. When the output is 1, it is altered.
In an environment where there is interference signal with a similar frequency to the working radar signal, randomly altering the start points of ramp-up time can make the mixer’s output stand out of the IF bandwidth and then be filtered out, and achieves similar effects as in Frequency Hopping mode.
In this mode, initial phases of chirps inside a frame are randomly altered by the XOR chain as shown in Fig. 5. When an interference signal appears, due to the randomization of phases, its energy will be distributed in the spectrum of the entire 2D-FFT and as a result no false target will occur.
In the above 3 modes, Alps SoC need make phase compensation for chirps in difference states so as to reduce phase error caused by chirp randomization. If chirps are not properly compensated, false targets are likely to occur in the velocity dimension of 2D-FFT spectrum. The proprietary baseband of Alps SoC can make compensation for the first 2 modes by default and for phase scrambling, can make more accurate compensation by calibrating the opposite phase with more accurate value, which is the best among the three as no spur occurs in the velocity dimension (Alps SDK now offers easy-to-use API for user calibration).
Apart from the 3 mentioned interference avoidance modes, baseband of Alps SoC also integrates interference mitigation algorithm. Upon receiving the time-domain signal waveforms, baseband will evaluate the changes in the amplitude of the signals. As shown in Fig.6, once samples with unusual amplitude changes are found, the signals will be deemed as interference. In this case, Alps SoC will mitigate the signals to reduce the noise floor of 2D-FFT.
To sum up, Alps SoC provides multiple features for users to address mutual interference among radar sensors, which prove to be effective in the suppression and mitigation of interference. For detailed algorithms and implementation, please refer to Alps Radar Baseband User Guide.