In our fourth webinar in the series about the DECT NR+ standard the deeper dive into technology continues. Heikki Berg from Nordic Semiconductor introduced the physical layer to the audience. He continued by explaining time and frequency synchronization, channel estimation and forward error correction & hybrid ARQ. The audience again had questions for the presenter, and we want to share the same information to all.  

Why is the preamble with numerology mu=1 equal to 14/9 symbols and for mu=2 it is more regular, i.e., equal to exactly 2 symbols?  

We are showing the capacity optimization compromise, so that for single slot transmissions there is additional payload OFDM symbol. 

Some countries allow DECT to operate only in 1910-1920 MHz, because of the WWAN networks in the vicinity. Is DECT NR+ able to operate on this reduced band?  

DECT NR+ can operate on a reduced band. Introduction to additional sub-bands to TR 103 636-2 to support this variability is being discussed. 

Why do the DRS symbols not cover the edge subcarriers when having single antenna? This complicates the interpolation process and CSI estimation for edge subcarriers.   

It actually simplifies it as the interpolation matrix is the same regardless of antenna. The frequency coherence is not dependent on absolute subcarrier position, but rather relative distances between subcarriers. A maximum of 3 subcarriers needs to be extrapolated. The same is necessary in the packet end as the channel estimates for the last OFDM symbols are extrapolated. 

How does the RX know the correct TBS? Is it directly taken from a table as in LTE and if so, why is there an option for filler bits before encoding?  

TBS is calculated from packet length and MCS (= Modulation and Coding Scheme). The TBS calculation has been implemented in a manner where the number of filler bits is always 0. 

Is there any control over the ARQ for higher layers to decide requests latency?   

Example scenario: optimize power consumption in sensor networks, assume DECT NR+ between gateways, with low data rate and very long latency accepted, avoid instant retry of the packets lost with default redundancy and FEC.  

Response window is agreed between radio devices, thus there is option to control power consumption. Currently higher layers do not have control over MCS selection, but this could be part of the radio profile. Higher layers can select if HARQ is used or not for transmission. 

Regarding synchronization: For good performance, is a simple autocorrelation of the received signal sufficient, or does one also need to cross-correlate against a precalculated template of the STF (= Synchronization Training Field)?  

For reliable performance close to sensitivity SNR (= Signal to Noise Ratio) ~ 0 dB cross correlation-based techniques are necessary. Autocorrelation based techniques work well only on positive, SNR > 6dB. Secondly, cross correlation-based techniques give more accurate timing estimates. Thirdly, as the STF is periodic, cross-correlation based techniques have lower computational complexity. 

Any idea of OFDM benefit in indoor vs other technologies? Have indoor tests also been done?  

Resilience against multipath propagation, especially in large indoor venues, for instance conference halls, stadiums etc. 

What are the performance criteria for the required SNR of 3.5dB?  

10% packet error rate for MCS1 is reached approximately at 3.5dB in AWGN. 

What FFT size is used in Dect NR+?  

The specification has FFT sizes 64-1024. However, operating at DECT band, the FFT sizes and subcarrier width selections which result to nominal bandwidth of 1.728, 3.456 or 6.912MHz, can be used. That is then FFT sizes 64 and 128 together with subcarrier widths of 27 a 54 kHz. The bandwidths supported are chipset implementation specific.  

Does the 10dB link budget improvement include the use of FEC and ARQ? Or do we get this from the OFDM alone? 

FEC mostly. If there is no FEC coding gain the packet error rate is directly dependent on the raw bit error rate of modulation. With GMSK transmission of 300 bits at 10% packet error rate requires raw bit error rate of 3.52e-4, which for GMSK transmission requires without any implementation losses requires 12 dB SNR. In practice however, higher SNR is required as classic DECT applications do not tolerate 10% packet error rate. 

If you are looking for more information on the DECT NR+ standard physical layer, go check out the webinar recording. 

We also urge you to read the previous blogs and watch the webinar recordings on the Dect NR+ standard: 

Introduction: FAQ Blog and webinar recording  

Applications and use cases: FAQ Blog IoT, FAQ Blog Audio and webinar recording 

Technology – upper layers: FAQ Blog and webinar recording