How to configure

Considerations

Start by asking yourself the following questions related to your application:

• What range do we need to cover, from \(r_\text{near}\) to \(r_\text{far}\)?

• Do we need to distinguish multiple objects? If so, how close could they be?

• Do we need to measure distance to objects? If so, what trueness is needed?

• If objects move, how fast could they go?

• Do we need to measure motions of objects? If so, at what speeds?

• With what rate do we need to obtain a result?

• What are our requirements on overall power consumption?

• What is the farthest distance from the sensor that an object may appear which could result in range ambiguities?

Range-related parameters

Having these questions and answers in mind, we go though the parameters in order, starting with those related to the range:

• Use the highest profile possible for maximum overall power and time efficiency. It is often limited by \(r_\text{near}\) since the close-in distance (CID) must be smaller than \(r_\text{near}\), and higher profiles have a larger CID. If you need to resolve multiple objects, the duration of the pulse must be short enough to give the resolution needed. Finally, lower profiles may give more precise distance measurements.

• The step length should be as long as possible to reduce memory usage and decrease measurement time. It is typically limited by two things:

  • Distance measurement trueness: As a rule of thumb, the steps need to be 1/10 to 1/2 of the required trueness. Note that as steps get smaller, other factors such as SNR and pulse duration (profile) have a bigger impact on the general accuracy.

  • The profile: If steps are too long, reflecting objects may fall between points, creating “blind spots” in the range. See Figure 22 for an example.

• From here, the start point and number of points can be set. Just make sure the points cover \(r_\text{near}\) to \(r_\text{far}\). Due to the pulse length (profile), the start and end points does not necessarily have to pass \(r_\text{near}\) and \(r_\text{far}\). Again, see Figure 22 for an example of this. However, keep in mind that distance measurements typically cannot be done in the very edge of the range, so you might have to extend it outside \(r_\text{near}\) and \(r_\text{far}\) anyways.

• Set the PRF such that the resulting maximum unambiguous range (MUR) extends beyond the farthest distance an object may appear. Keep it as high as possible since a higher PRF is more power and time efficient overall.

Figure 22 A sketch of setting up the measurement range for efficient coverage of a given area.

Rate-related parameters

With the range related parameters all set up, we move on to parameters related to sampling rate:

• If you need to estimate velocities, that typically means applying an FFT on a frame over sweeps to produce a distance-velocity (a.k.a. range-Doppler) map. In that case, the sweeps per frame (SPF) sets the frequency (velocity) resolution. For e.g. micro- and macro gesture recognition, typical values range from 16 to 64. For more accurate velocity measurements, typical values are much higher ranging from 128 to 2048.

  If you don’t need to estimate velocities but still need to detect “fast” motions (\(\gtrapprox 500 \text{Hz}\)), that typically means estimating the energy within a frame. For such cases, e.g. running, walking, waving, typical SPF:s range from 8 to 16.

  If you need to detect “slow” motions or have a mostly static environment, there is no need to use multiple sweeps per frame (SPF), so set it to 1. Such cases include (inter-frame) presence detection and distance measurements.

• For cases where SPF = 1, the sweep rate is not applicable. What matters then is setting the HWAAS to achieve the needed SNR. Keep in mind that measurement time linearly increases with HWAAS, so keeping it as low as possible is important to manage the overall power consumption.

  For cases where SPF > 1, the sweep rate \(f_s\) should be adapted to the range of speeds of interest. A good rule of thumb is

  (12)\[f_s \approx \frac{10}{\lambda_{RF}} \cdot |v|_\text{max} \approx 2000 \text{m}^{-1} \cdot |v|_\text{max}\]

  where \(|v|_\text{max}\) is the possible maximum relative speed between the radar and the object (in m/s).

  In many cases, the most efficient way to achieve the target sweep rate is by adapting the HWAAS. The sweep rate is inversely proportional to the number of HWAAS. It is also possible to directly control the sweep rate, letting the sensor idle between sweeps. However, idling between sweeps is rarely as efficient as idling between frames.

• If power consumption is not an issue, start by using the highest possible frame rate. Otherwise, it is crucial to minimize the frame rate to let the sensor idle in a lower idle state as much as possible. See Table 8 for typical values.

Table 8 Typical parameter values for some applications.

Application	Frame rate
Micro gesture recognition	30 - 50 Hz
Medium power presence detection	10 - 80 Hz
Low power presence detection	1 - 5 Hz

Idle-related parameters and control

Two idle states are available to trade off power consumption during idle periods against measurement setup time:

• READY - The deepest idle state. As much of the radar subsystem as possible is shut down.

• ARMED - The shallowest idle state. Most of the radar subsystem is kept on.

Because ARMED keeps most of the radar subsystem active, the time to transition from ARMED into a measurement is significantly shorter than from READY.

The idle state is controlled by two parameters:

• inter sweep idle state - the idle state the sensor enters between each sweep within a frame.

• inter frame idle state - the idle state the sensor enters between frames.

The default value for both parameters is ARMED.

If inter sweep idle state is set to READY, then inter frame idle state must also be set to READY.

As a general rule, set both idle states to the deepest state that still allows the target sweep and frame rates to be maintained. Use ARMED when low measurement latency or high sweep and frame rates are required, since transitioning from ARMED is significantly faster than from READY. Use READY when minimizing power consumption during idle periods takes priority and the longer transition time is acceptable.

Channel configuration

The A212 has 3 transmitter (TX) and 4 receiver (RX) antennas. By selecting which antennas are active, the measurement is performed over a set of virtual channels, each corresponding to one active TX and RX antenna pair. The number of virtual channels is the product of the number of active TX and RX antennas. Activating all 3 TX and all 4 RX antennas gives 12 virtual channels, the maximum for the A212.

The active antennas are controlled per subsweep through two bit mask parameters:

• tx antenna mask - selects active TX antennas. Bit 0 maps to TX0, bit 1 to TX1, bit 2 to TX2. Setting all three bits (0b111 = 7) activates all TX antennas (default). At least one bit must be set.

• rx antenna mask - selects active RX antennas. Bit 0 maps to RX0, bit 1 to RX1, bit 2 to RX2, bit 3 to RX3. Setting all four bits (0b1111 = 15) activates all RX antennas (default). At least one bit must be set.

Example 1: single channel (TX0 + RX0 only): Set tx_antenna_mask = 0b001 = 1 (bit 0 set, TX0 active) and rx_antenna_mask = 0b0001 = 1 (bit 0 set, RX0 active). This yields 1 × 1 = 1 virtual channel (TX0 and RX0 pair).

Example 2: six channels (TX0, TX1 with RX0, RX1, RX2): Set tx_antenna_mask = 0b011 = 3 (bits 0 and 1 set, TX0 and TX1 active) and rx_antenna_mask = 0b0111 = 7 (bits 0, 1, and 2 set, RX0, RX1, and RX2 active). This yields 2 × 3 = 6 virtual channels (all TX0/TX1 and RX0/RX1/RX2 pairs).

Different subsweeps within the same sensor configuration may use different antenna masks, allowing flexible multi-channel configurations within a single frame. See Frames, sweeps and subsweeps for details on how channels appear in the output data.

Other parameters

• Leave receiver gain at the default value and reduce if saturation occurs.

• Leave enable TX set (default).

• Leave jitter compensation enabled (default) for improved phase stability between sweeps. It occupies one of the five subsweep slots, so a configuration using all five has to disable it. Disable it also when the highest possible sweep rate is required.