Unified Air Interfaces in 5G

With the development of wireless communications, a consensus has been reached that the future 5G system should be a unified network adaptable to different scenarios.

5G scenarios can be divided into three groups: mobile broadband (MBB); massive connection (mMTC: Massive MTC); and high reliability, low-latency communication (cMTC: Critical MTC). MBB provides high capacity and high speed, but it is relatively insensitive to the number of connections and reliability. mMTC provides energy efficient, low-cost access for a mass number of nodes, but the data transfer rate is not high. cMTC is mainly designed to decrease delay and increase the reliability of data transmission, but it is not designed to accommodate a mass number of nodes or provide a high data transmission rate. Compared with previous generations of WANs, the 5G access network needs to be flexible and open. It must be adaptable to individual demands and provide external standardized interfaces that enable users to accomplish specific tasks through the 5G access network platform. Therefore, the priority is designing unified air interfaces so that the 5G access network can efficiently support different services.

The structure of 3G and 4G air interface protocol stacks mainly supports mobile broadband data transmission (of which voice transmission can be considered a part). However, this single structure cannot meet the requirements of different services provided by 5G air interfaces. Tests have proven that 3G and 4G air interface protocol stacks support mobile broadband data transmission. The design of the 5G unified air interface protocol stack should be based on this, properly expanding on the 3G and 4G air interface protocol stacks.

The 5G unified air interface protocol stack introduces the L1 layer, which is the abstract physical layer. This layer is designed to extract common points of different services on the physical layer. Common points are not necessarily exactly the same, but they can be configured to be the same. Therefore, the L1 layer is transparent to various services and frequency bands. Currently, the identifiable contents of the L1 layer include waveform and frame structure parameters. CP-OFDM has been widely used in LTE, so the selected 5G waveform should be able to coexist with CP-OFDM very well. That is, CP-OFDM needs to be changed so that the 5G waveform can suit some scenarios, e.g., in the case of low out-of-band leakage or low time- and frequency-domain synchronization. A good waveform is FB-OFDM, which filters CP-OFDM at the sub-carrier level so an efficient polyphase filter can be designed. Also, because each sub-carrier filters the out-of-band leakage, the waveform has low out-of-band leakage and is robust in terms of frequency domain synchronization. The CP or stretched symbols after polyphase filtering are more robust to multipath radio channels. Generally, the only difference between CP-OFDM and FB-OFDM is that FB-OFDM has a polyphase filter. If the polyphase filter is defined as one beat, then FB-OFDM can roll back to CP-OFDM smoothly. In the 5G system, different services are carried on different frequency bands, so the frame structure parameters should be different. It is not good for each frequency band to have independent frame parameters. An efficient and flexible method involves scalability. Taking the frame structure parameters in LTE as a starting point, we define a scalable factor S. All other parameters, such as sampling frequency, sub-carrier interval, symbol length and CP length, are controlled by this parameter. As long as the scalable parameter is configured properly, different services and frequency bands can be supported by the frame structure. For example, we generally configure higher scalable parameters at a high frequency to support larger bandwidth, using a larger sub-carrier interval, shorter symbol length, shorter CP length and shorter TTI.

In terms of the slice design on L1, L2 and L3, different services have different demands, so these layers need to be designed accordingly. For MBB service, L1 focuses on Massive MIMO, SVC, and high-frequency beam tracing. In addition, MBB service has abundant sub-services, so it requires an entire L1/L2/L3 protocol stack structure. For mMTC service, the L1 layer mainly focuses on supporting mass access, so the non-orthogonal access mechanism MUSA is a good choice. MUSA can multiplex connections more than three times on a single time-frequency, so it has an overload rate of more than 300% and low complexity. Compared with MBB, mMTC has less service data, so its L2/L3 protocol stack is not exactly the same as that of MBB. Reconstruction and consolidation are needed to reduce the overhead of the protocol stack. Low latency is required so that the protocol stack is simplified as much as possible. In this way, multiple channels are terminated at L1 and do not require a complete L2/L3 protocol stack. Although high reliability involves diversity and redundancy at different levels, L1, L2 and L3 have different characteristics. For example, multi-connection is defined on L3; error correction codes are defined on L2; and frequency, time and space diversity are used on L1.

The 5G unified air interface protocol stack also introduces the L3+ layer, which is the service-perception layer. Traditional access networks generally do not define services and have weak mechanisms for controlling data flow. However, in the 5G system, the access network has to support different services and frequency bands, and it is necessary to introduce a service-perception layer. This layer is designed to carry on the bearer from the core network to the access network and distinguish different services in the access network. In this way, each service is carried on a different slice and configured with corresponding transmission parameters on L1.

The introduction of a carrier-class operating system has been a long-term goal for the 5G unified air interface protocol stack. To make the access network more open, it necessary to build a carrier-class operating system platform that is connected to each of the layers of the 5G unified air interfaces. Moreover, standard APIs are provided for operators, equipment manufacturers, third-party developers, and even individuals to develop and customize on-demand.

The above describes several aspects of the unified air interface design of 5G access network from the perspective of protocol stack. Unified, flexible, open 5G air interfaces are possible. 5G unified air interfaces provide a unified access mechanism for various services and frequency bands and meet the long-term requirements of operators and customers for future 5G networks.