Content for  TR 38.877  Word version:  18.1.0

In this section, a number of other considerations for FR2-1 wideband BS from the feasibility point of view that are not considered in the preceding sections are mentioned. Digital considerations To support a large bandwidth, it is necessary to provide a high sampling rate for ADC and DAC. For individual Frequency Band Groups (e.g. 24.5-29.5GHz, n257+n258), the bandwidth to be supported may be more than 5GHz, and to cover the full FR2-1 the bandwidth might be up to 24GHz. If linearization would be seen later to be feasible for multiband BS, then DAC/ADC would need to support for the feedback and correction, which poses another challenge to DAC/ADC. Potentially only the in band spectrum could be generated by separate converters which would reduce the requirements on individual converters. In an implementation, the architecture, RF performance and power consumption of the analogue/digital interface would be key considerations. The ADC and DAC complexity and power consumption could be reduced by reducing the sampling resolution, but this would impact TX factors such as EVM and emissions and RX factors such as dynamic range and RX EVM. For this reason, care would need to be taken that requirements on e.g. EVM, receiver dynamic range and demodulation performance would be achievable. In addition to the DAC and ADC, the large bandwidth and sampling rates would lead to very high volumes of data to be moved within the radio architecture. Data transport and interface architectures would need to support the very high data volume. Reducing the data volume (e.g. by reducing the sampling resolution) would lead to similar considerations as for ADC and DAC on meeting requirements such as EVM, emissions, RX dynamic range and demodulation. In addition to the interface bandwidth, the digital transport latency may also impact radio near algorithms (such as DPD, CFR) and could impact the performance of the transmitter and receiver. It is not clear whether the need to support a much larger interface bandwidth could impact the interface latency. Digital filtering may be needed for meeting selectivity and blocking requirements depending on the sensitivity and architecture. The large sampling rates and bandwidth would increase the amount of computational power needed for digital filtering, and potentially the achievable steepness of the filters. This could impact the feasibility of meeting TX EVM and RX selectivity, blocking and demodulation requirements. Analogue considerations The possibilities for analogue filtering within an FR2-1 AAS array are extremely limited. The filter in a typical single-band FR2 BS may not be placed before the antenna. However, it could be the case of multi-band, due to the multi-frequency signal going through the PA if the PA is highly nonlinearity, or due to the architecture splitting multi-band signals. And thus filter before the antenna may be needed. Choosing where to place filter components is an important part of a multi-band solution as it impacts the overall performance. For instance, if a filter would need to be placed before the antenna, then there are other issues in practice with insertion loss and power unbalancing due to non-identical filter for each branch. The regulatory requirements in the inter RF gap and the linearity of the transmitter will set the boundary for the analogue filtering. The feasibility of analogue filtering may impact the ability to meet some regulatory emissions requirements and out of band blocking, and if the filter has ripple also the EVM may be impacted. Depending on the array architecture, it may be necessary to split different frequency components of the multi-band signal and route them to different antenna elements, e.g. in architectures using diplexer. The splitting and additional routing can have implications for TX power loss and RX sensitivity, and the placement of components needs to be carefully considered. For instance, placing the diplexer before the PA is preferred from a design loss and power dissipation perspective. This is because the insertion loss of the diplexer occurs at a lower power level and can be accommodated by a simple increase in driver gain. Conversely, if the diplexer is placed after the PA, the insertion loss of the diplexer directly reduces the power available to the antenna array and leads to more thermal dissipation. On the other hand, it is worth noting that if the power loss is significantly different/uneven between antenna branches, then beamforming may be degraded, and calibration may be required which may bring complexity to the architecture. This may impact the feasibility of the multi-band solution, although since TX power and RX sensitivity are subject to declarations it may not impact the requirements definition. As shown in Clause 4.2.1 Option #4, splitting and additional routing can be avoided by using a multi-band antenna, such as a single input stacked patch design. Dedicated PAs for the desired frequency ranges may be adopted from a design loss and power dissipation perspective, as the narrower band designs will have better efficiency and the thermal dissipation will be spread out over a larger area.