Anomalous distances

[Vorstok]

Hello, has anyone encountered any abnormal measurements? I was trying to track down their cause and it turns out that the TOF is not being measured correctly. The computeRangeAsymmetric function in DW1000Ranging.cpp calculates myToF using four timestamps

void DW1000RangingClass::computeRangeAsymmetric(DW1000Device* myDistantDevice, DW1000Time* myTOF) {
    // asymmetric two-way ranging (more computation intense, less error prone)
    DW1000Time round1 = (myDistantDevice->timePollAckReceived - myDistantDevice->timePollSent).wrap();
    DW1000Time reply1 = (myDistantDevice->timePollAckSent - myDistantDevice->timePollReceived).wrap();
    DW1000Time round2 = (myDistantDevice->timeRangeReceived - myDistantDevice->timePollAckSent).wrap();
    DW1000Time reply2 = (myDistantDevice->timeRangeSent - myDistantDevice->timePollAckReceived).wrap();
    myTOF->setTimestamp((round1 * round2 - reply1 * reply2) / (round1 + round2 + reply1 + reply2));
}

The error occurs mainly due to incorrect measurement of round1

I know that anchors are polled sequentially. But I don't understand what to do next. Can you please advise me who has faced this problem and how to solve it. Below I attach two screenshots with measuring distance of 4 anchors in motion and in static position.

[Vorstok]

In the firmware for Tag esp32 in link.cpp there is a filter with a moving average. It creates damped bursts with anomalous data. I commented out the filter. And the spikes became easy to remove by a sharp jump between the new and old value.

before:

after:

ps: Anchor 85 this is stm32