Content for  TR 38.844  Word version:  18.0.0

A typical UE architecture utilises a number of filters of two major types - analogue and digital - and it is generally up to the UE implementation how they are combined. Nevertheless, it is often the case that a UE uses first the wideband analogue filter which typically covers a whole band. In addition to that, a UE may use another NR channel bandwidth specific analogue filter, premise function on which is to filter our non-adjacent blockers. However, since even the NR channel bandwidth specific analogue filter cannot ensure "brick-wall" like filtering, a UE also applies digital filter after ADC to eliminate adjacent interferer. Depending on the UE implementation, the digital filter is a combination of the hardware and software components that allow the UE to apply the corresponding filter coefficients to support a wide range of standard channels, e.g. from 5MHz to 100MHz in case of FR1. As an example, consider Figure 6.1.3-1. In this example, RBs are scheduled close to the NR band edge. The potential adjacent interferer below the band edge will be removed by the CBW filter with no degradation of ACS at the lower side of the irregular BW since the filters have the same capability as the regular CBW operation.

However, for the scenario in the Figure 6.1.3-2 the wider channel filter cannot protect against adjacent interferer(s) when the irregular spectrum block is narrower than the channel filter and/or there are adjacent interferers on both sides. Current specifications do not define how a UE configures its digital filter within the configured channel bandwidth. So, in the provided example below, a UE implementation may configure the digital filter in accordance with the carrier bandwidth "ignoring" the actual smaller bandwidth part size. This is illustrated further in Figure 6.1.3-2. The wanted signal is smaller than 10MHz, but the UE filter is always set to 10MHz as signalled by the network. As can be seen, if there is an adjacent interferer, then it can "leak" into the wanted signal region.

To estimate anticipated performance described in Figure 6.1.3-2 when a UE configures its digital filter according to the channel bandwidth potentially allowing the adjacent interferer to "leak" into the wanted signal area, a series of measurements were conducted using the same testbed and setup as used for the commercial device conformance testing. The adjacent interfering signal is always set in such a way that it is right next to the wanted irregular channel bandwidth signal. Table 6.1.3-1 below summarises adjacent interfering channel parameters used in measurements, where adjacent interferer offset is the offset from the wanted signal center to the adjacent interfering signal center.

The following common parameters were applied:

• Center frequency: 1850MHz
• UE channel: 10MHz
• Wanted signal level at UE antenna: -82dBm
• Adjacent interferer level at UE antenna: from -120dBm to -50dBm in step of 1dB
• Adjacent interferer: 5MHz, 16QAM, SCS 15kHz (3GPP standard ACS test)

Intermediate irregular channel bandwidths are considered - from 9MHz to 6MHz - and the adjacent interferer offset is set accordingly. As the main performance indicator, the resulting SNR is estimated over the whole 10MHz region irrespective of the actual irregular channel size. This approach provides the worst-case estimation of the resulting SNR because in the real life a UE will most likely estimate SNR over the configured bandwidth part corresponding to the irregular channel size or its sub-part. As can be seen from Figure 6.1.3-3 below, when the adjacent interferer level is low then the estimated SNR is around 39dB for all irregular channels (if the wanted signal is -82dBm and the adjacent interferer is -120dBm, then the theoretical SNR would be around 38dB, but in our case the adjacent interferer is not even fully inside the wanted signal bandwidth). The 9MHz irregular channel bandwidth has same SNR of 39dB even when the adjacent interferer level is same as the wanted signal. SNR degradation starts at the adjacent interferer level of approximately -72dBm, i.e. 10dB higher than the wanted signal. When the adjacent interferer level is 32dB higher than the wanted signal, i.e. -50dBm, the estimated SNR is around 27dB for the 9MHz irregular channel. As for the opposite extreme case of the 6MHz irregular channel, SNR degradation can be already observed at the adjacent interferer level of around -100dBm, i.e. 18dB lower than the wanted signal. When the adjacent interferer level becomes as high as the wanted signal, the estimated SNR drops down to 27dB. Finally, when the adjacent interferer is 32dB higher than the wanted signal, the SNR becomes -5dB. These results show the relative degradation in SNR that can occur due to adjacent channel interferers leaking through the next larger BW filter, and the actual performance degradation will depend on a particular scenario, especially ACI strength and the UE channel filter overlap with the interfering signal. Typical 3GPP ACS requirements are conducted so that a UE should be able to meet 95% of the target throughput of the reference QPSK level when the adjacent interferer is 32dB higher than the wanted signal (and which corresponds to approximately -1.5dB SNR). Referring to the right-most point on the X-axis of Figure 6.1.3-3, it can be seen that 9MHz and 8MHz irregular channel SNR is above -1.5dB, while the 6MHz channel SNR is clearly below -1.5dB.

To further reduce the risk of ACS reduction, as shown in this second example in Figure 6.1.3-2 and Figure 6.1.3-3, it is possible for new UEs to align their digital filter to the actual smaller bandwidth part within the wider analogue CBW filter bandwidth.