nRF9151 DECT NR+: Splitting devices into frequency groups using different carriers — feasibility, RX filtering, and minimum channel spacing

[Mars Hsu]

I am building a multi-device walkie-talkie system using DECT NR+ on the nRF9151. I want to split devices into isolated communication groups by assigning each group a different carrier (channel), and I have a few questions about how this works at the PHY layer.

Background / what I understand so far:

I understand that "band" refers to the regulatory frequency range (e.g. Band 1 for Europe, Band 9 for North America), and within a band there are multiple carriers spaced approximately 1.728 MHz apart, selectable via nrf_modem_dect_phy_tx_params.carrier per TX operation. This is not automatic frequency hopping — the carrier is explicitly chosen by the application.

My intended design: I have N devices all operating within the same band. I want to assign group A (e.g. 3 devices) to carrier X, and group B (e.g. 2 devices) to carrier Y, so that the two groups communicate independently without interfering with each other. Group isolation is managed entirely in the application layer — devices in group A only TX and RX on carrier X, and vice versa.

Questions:

1. RX carrier filtering — does the PHY filter out packets on non-matching carriers automatically? When I schedule an RX operation on carrier X, will the modem hardware reject / not report packets that arrive on carrier Y? Or will the RX callback fire for any received packet regardless of carrier, and my application needs to discard non-matching ones?

2. Minimum carrier spacing for isolation between groups Is one carrier step (~1.728 MHz) sufficient to ensure that group A and group B do not interfere with each other, even when all devices are in close physical proximity? Is there a recommended minimum number of carrier steps between the two groups for reliable isolation?

3. Any known caveats with this multi-carrier group design at the PHY layer? Are there any PHY-level restrictions, scheduler constraints, or known issues when multiple groups of devices operate on different carriers simultaneously in the same physical space? For example, does the LBT (Listen Before Talk) mechanism interact with transmissions on adjacent carriers?

Hardware: nRF9151 Connect Kit (Makerdiary), board target nrf9151_connectkit/nrf9151/ns NCS: v2.9.0-rc1-767-g121a3d3a76 (nrf pinned to 98a5e50b9ac142ae2114f0b4ea257823cbf2c978) Modem firmware: [v2.0.0]

[Simon D-M]

Hi,

I have some guess for your question, but I don't want to write them now as I am not very familiar with DECT NR+. Let me just confirm them internally with the DECT team to see if they are correct or not.

I'll send you a better response as soon as my guess have be confirmed or corrected (before the end of the week).

Sorry for the delay,

Best regards,

Simon D-M

[Simon D-M]

Hi,

I got some answers from the DECT team:

1. RX carrier filtering — does the PHY filter out packets on non-matching carriers automatically? When I schedule an RX operation on carrier X, will the modem hardware reject / not report packets that arrive on carrier Y? Or will the RX callback fire for any received packet regardless of carrier, and my application needs to discard non-matching ones?

The PHY can and will only operate on one channel at a time. Rx on carrier "X" will not receive anything from carrier "Y".

2. Minimum carrier spacing for isolation between groups Is one carrier step (~1.728 MHz) sufficient to ensure that group A and group B do not interfere with each other, even when all devices are in close physical proximity? Is there a recommended minimum number of carrier steps between the two groups for reliable isolation?

It depends quite a lot on the used band. There are links to local regulation in the "important" box in this documentation (link). Also worth mentioning that the radio specification requirement for adjacent channel selectivity is around 25dBs. Currently, our radio implementation has around a 5dB margin to specification. Because of that if device A transmits on an adjacent channel near device B, it will interfere with device B reception if the adjacent channel power at device B antenna is about 30 dB higher than the wanted signal.

So if the devices are quite close to each other, it would be wiser to have more separation in the frequency domain between the 2 groups. (If the carrier separation is doubled to 3.456MHz, the isolation increases from ~30 dBs to over 70dBs.)

3. Any known caveats with this multi-carrier group design at the PHY layer? Are there any PHY-level restrictions, scheduler constraints, or known issues when multiple groups of devices operate on different carriers simultaneously in the same physical space? For example, does the LBT (Listen Before Talk) mechanism interact with transmissions on adjacent carriers?

Even though this is more of a MAC/Application level concern as this is not handled in the PHY directly, but please be careful to follow the regional regulations on channel access before transmitting.

Also on the LBT mechanism, the specification limit for adjacent channel leakage ratio is -30dBc. Thus, a device transmitting at +19dBm will leak a maximum of 11dBm power to the adjacent channel. Depending on how much isolation there is between the channels, there might be enough power leakage to trigger the LBT.

For example, according to the Friis transmission formula, the free space loss at 1.9 GHz for 10 meter is 58dBs. Which means that the adjacent channel power leakage at the other device antenna will be around -69dBm (-11dBm - 58dB). According to the specification, LBT threshold for band 1 device with 19dBm output must be set to -71dBm or lower, which would trigger the LBT. 

I hope I was able to help you. If something isn't clear, feel free to ask for clarifications.

Best regards,

Simon D-M