5G New Radio

5th Generation Mobile Network or simply 5G is the forthcoming revolution of mobile technology. Basically, 5G is the advanced form of the LTE-A Network. In the 5G Network may be used advanced form key technology such as MIMO, OFDM, SC-FDMA etc.

Ultra-dense small cells are foreseen to play an essential role in the 5th generation (5G) of mobile radio access technology, which will be operating over different bands with respect to established systems. The natural step for exploring new spectrum is to look into the centimeter-wave bands as well as exploring millimeter-wave bands. This article presents our vision on the technology components for a 5G waveforms concept for ultra-dense small cells. Fundamental features such as optimized short frame structure, multi-antenna technologies, interference rejection, rank adaptation and dynamic scheduling of uplink/downlink transmission are discussed, along with the design of a novel flexible waveform and energy-saving enablers.

We have been adopted new physical waveforms or new modulation scheme at least. The new waveform in a new generation has been designed to overcome some restriction in previous technology or increase the spectrum efficiency. It is highly likely that this holds true for 5G, implying that we will see the new waveform.

1.1. UW-DFT-S-OFDM

The transmitter structure for UW-DFT-S-OFDM (Unique Word Discrete Fourier Transform Spread Orthogonal Frequency Division Multiplexing) is a small modification to the well-known DFT-S-OFDM, CP-less DFT-S-OFDM can be generated by adding a unique word to the input of the DFT operation,

At the receiver, DFT-FDE-IDFT modules in LTE-SC-FDMA can be used for UW-DFT-S-OFDM and CP removal module is not needed. Insertion of the unique word prior to the DFT operation leads to suppression of out-of-band emission, thanks to the cyclic property obtained from the DFT and IDFT operation. Location of UW is an implementation issue and depends on the target level of OoB emission and desired protection against ISI when going through a multipath channel since UW plays the role of CP. 

UW sequence is placed as shown in Figure. In the figure, the number of symbols in UW and the number of allocated sub-carriers are denoted by M and ND, respectively. As shown in the figure, UW is split in half and placed at the head and tail of the DFT-S-OFDM block.

Conclusion

• UW DFT-S-OFDM has superior OoB suppression performance with respect to SC-FDMA.
• UW-DFT-S-OFDM and SC-FDMA have the same PAPR performance.
• UW serves as CP and sufficient FER performance can be obtained by controlling length of UW
• UW-DFT-S-OFDM is robust against interference from adjacent channels
• UW-DFT-S-OFDM is a waveform with suppressed OoB, robust performance in multipath fading and tolerance against adjacent channel interference.

1.2 Universal Filtered Multi-Carrier

UFMC is also based on the filtering method. The differentiator is how and what type of the filter is applied. In the case of FBMC, it packs multiple consecutive subcarriers into a group called 'subband' and applies the filter onto each of the sub-band as illustrated below. How many sub-carries would belong to one subband is a design specification. The more sub-carries in one subband, the less load on baseband processing, but final performance gets poorer as you put more sub-carriers into one subband.

We are going to filter a subgroup of carriers, so multiple carriers in a bank filtered together. This give us the ability to improve the out of band emission and it gives us the ability to improve the spectral efficiency and at the same time we can have cyclic prefix if we want, so we can still support MIMO in this filter technology.

 Overall tranciever structure of FBMC can be illustrated as below . As you see from the top left, you see multiple subcarriers are grouped into a subband. For example, s_1k indicates subband 1 that is composed of k subcarriers. And each subband goes to a separate IDFT process and converted into Time Domain sequence and then get filtered. Then, output sequence of each filters are combined and upconverted to RF and get transmitter.

The effect of applying filters to a single subband can be obvious comparing to OFDM as shown below.You would notice much-improved stop-band rejection in UFMC (green) comparing to the OFDM.

In real use case, it would be common to allocate the multiple subband sitting next two each other as shown below. Since the rejection of stopband has been improved a lot, the interference between a subband and neighbouring subband can be much lower comparing to OFDM case

1.3 Orthogonal Time-Frequency & Space

The new 2D modulation technique is called OTFS (Orthogonal Time Frequency & Space). The OTFS transforms information carried in the Delay-Doppler coordinate system to the familiar time-frequency domain utilized by traditional modulation schemes such as OFDM, CDMA and TDMA. OTFS converts the fading, time-varying wireless channel into a non-fading, time-independent interaction with the transmitted symbols, revealing the underlying geometry of the wireless channel. In this new formulation, all QAM symbols experience the same channel and all Delay-Doppler diversity branches of the channel are coherently combined. Because channel state acquisition is done in the time independent Delay-Doppler domain, accurate channel estimation is achieved, even in the presence of high-mobility. In addition, because antenna port reference signals are carried in the Delay-Doppler domain they can be packed very efficiently, allowing large numbers of reference signals to be flexibly multiplexed based on the delay and Doppler spread characteristics of the individual channels.

OTFS Modulation Overview

OTFS works in the Delay-Doppler coordinate system using a set of basis functions orthogonal to both time and frequency shifts. Both data and reference signals and pilots are carried in this coordinate system. The Delay-Doppler domain mirrors the geometry of the wireless channel, which changes far more slowly than the phase changes experienced in the rapidly varying time-frequency domain. OTFS symbols experience the full diversity of the channel over time and frequency, trading latency for performance in high Doppler scenarios.

The figure illustrates the modulation and demodulation steps. The transmit information symbols (QAM symbols) are placed on a lattice or grid in the two-dimensional Delay-Doppler domain and transformed to the time-frequency domain through a two-dimensional Symplectic Fourier Transform. Through this transform, each QAM symbol is spread throughout the Time-Frequency Plane (i.e., across the selected signal bandwidth and symbol time) utilizing a different basis function. As a result, all symbols of the same power have the same SNR and experience exactly the same channel. The implication is that, given the appropriate frequency and time observation window, there is no frequency or time selective fading of QAM symbols. The transform converts the multiplicative action of the channel into a 2D convolution interaction with the transmitted QAM symbols. OTFS allows for the same OFDM shaping benefits seen in various forms of filtered OFDM. OTFS extracts the full diversity of the channel at the modulation level, allowing the FEC layer to operate on a signal with uniform Gaussian noise pattern, regardless of the particular channel structure... OTFS enables a flexible trade-off between observation time or latency for increased performance in high Doppler scenarios. For non-latency-sensitive traffic, such as video in a high-speed scenario, this is a reasonable trade-off. For more latency-sensitive scenarios, OTFS allows scaling of the observation to a single OFDM symbol.

Conclusions:- OTFS is a new 2D air interface paradigm with important spectral efficiency advantages in high-order MIMO and high Doppler scenarios, reference signal efficiency and channel estimation and prediction. All reference signals and QAM symbols are carried in the delay-Doppler domain and experience the same channel response over the transmission/observation interval and extract the maximum diversity of the channel in both time and frequency dimensions. This allows the FEC layer to operate on a signal with uniform Gaussian noise pattern, regardless of the particular channel structure. OTFS has a natural architectural compatibility with OFDM, based on its underlying multicarrier components and the reference signal architecture supports any form of multicarrier modulation. 

3GPP has identified a variety of eMBB deployment scenarios that focus on high vehicle speed and massive MIMO antenna arrays. The new radio air interface must support high spectral efficiency in high Doppler environments while supporting a large number of antennas. OTFS is ideally suited for these requirements, providing: high spectral efficiency; accurate channel estimation and prediction; and very efficient and flexible reference signals for massive MIMO applications.

1.4 Generalized Frequency Division Multiplexing 

GFDM is based on traditional filter bank multi-branch multi- carrier concepts which are now implemented digitally. Our GFDM approach exhibits some attractive features which are of particular importance for scenarios exhibiting high degrees of spectrum fragmentation. Spectrum fragmentation is a typical technical challenge of digital dividend use cases, exploiting spectrum white spaces in the TV UHF bands which are located in close proximity to allocated spectrum. Specifically, the GFDM features are a lower PAPR compared to OFDM, a ultra-low out-of- band radiation due adjustable Tx-filtering and last but not least a block-based transmission using cyclic prefix insertion and efficient FFT-based equalization. GFDM enables frequency and time domain multi-user scheduling comparable to OFDM and provides an efficient alternative for white space aggregation even in heavily fragmented spectrum regions.

GFDM is a flexible multi-carrier transmission technique which bears many similarities to OFDM. The main difference is that the carriers are not orthogonal to each other. GFDM provides better control of the out-of-band emissions and the reduces the peak to average power ratio, PAPR. Both of these issues are the major drawbacks of OFDM technology.

 In this scheme, a filter called Pulse Shaping Filter is applied per each sub carrier and multiple symbols per sub carrier are processed in a single step. (In this illustration, M indicates the number of symbols and K indicates the number of subcarriers

GFDM Frame structure can be compared to the current LTE structure as shown below. (The block diagram shown above may become clearer to you if you grasp the image of this frame structure first). As you see here, GFDM frame will be very short comparing to current LTE OFDM symbol to meet 5G latency requirement.

1.5Filter-bank based multi-carrier

In FBMC, Each of the individual sub-channel is filtered on its own. It uses the very narrow band filter with long time length. This gives us very good control of each filter bank. It gives a good control over emission of each of the subcarriers. We have very good spectral efficiency and very good data rate but because we don't have cyclic prefix, it is a little difficult to process MIMO. It is not supporting very well for the large MIMO array technology. For those application with MIMO or short time/burst transmission, it is not so effective.

The important thing is that each of the sub-carrier goes through a filter called 'Pulse Shaping Filter'.

The process illustrated above can be represented in a more intuitive way as shown below. As you see, each of sub-channel goes through a bandpass filter.

The critical steps for FBMC are to implement filters for each sub-channels and align the multiple filters into a filter bank. The way to build the filter bank is .. we design a basic form (template) of a filter called prototype filter. Once we finish the design of the prototype filter, the next step is simple. Just make a copy of the prototype filter and shift it to neighbouring sub-channels step by step.

1.6 Filtered-OFDM

F-OFDM stands for 'Filtered OFDM'. This term has confused me so much. Basically, most of 5G waveform candidates use a kind of filter. The difference among them is what kind of filter is used and how the filtering applies. The word 'Filtered' in F-OFDM doesn't give me any special meaning. Anyway, in some case, name is not so important... the important thing is the real meaning/contents behind the name.

 The real meaning behind F-OFDM can be illustrated as below. As you see here, in F-OFDM.. a band can be devided into multiple subbands. Each of the bands can have different bandwidth. Another point you should notice that each subband is made up of multiple subcarriers and the frequency spacing between the subcarriers can differ with each subband. Combining this subband flexibility and subcarrier flexibility, you can create very flexible structure of subframe that can carries the different types of service data within the same subframe. Based on the subframe requirement for 5G, it is likely that this kind of flexible waveform will be adopted in 5G.

A more detailed illustration showing the fundamental difference between F-OFDM and conventional OFDM can be illustrated. As you see here, in conventional OFDM the whole band is made up of a single block and the frequency spacing between each subcarrier are all same. In contrast, in F-OFDM, the whole band is made up of multiple subbands). The subcarrier spacing in each subband are different (e.g, the subcarrier spacing for N1 subband is delta_f/2 and the subcarrier spacing for Nk is 4*delta_f) and each subband has its own CP (the length of each CP may vary as well) and each subband is applied by its own filter.Obviously, the main advantage of this waveform would be flexibility and the main disadvantage would be the complexity of structure and implementation.

1.7. Windowed-OFDM

The OFDM signal spectrum should be shaped prior to transmission to lower the out of band emissions. The OFDM spectrum roll-off can be controlled at baseband by using the windowing and overlapping of consecutive OFDM symbols, as illustrated in Figure. This concept is not alien to LTE systems and is described in TR 25.892. Windowing eases out the abrupt phase differences at the OFDM symbol boundaries and brings down the out of band emissions. It does come at the cost of time-domain overheads, however, the window length W can be absorbed in the CP duration, leading to no extra overheads.

OFDM symbol are applied to the time window, our technique utilizes only a set of sub-carriers located at the edge of the band for windowing. While the size of the cyclic extensions for the non-windowed sub-carriers remains, the sizes of the cyclic ex- tensions for the windowed sub-carriers are reduced. Therefore achieving both spectrum efficiency and side lobe suppression.Rectangular windowing of OFDM symbols produces high side lobes, which results in adjacent channel interference (ACI). It would be desirable to reduce the ACI while maintaining a high level of spectrum efficiency for OFDM based systems.

1.8 Zero-tail Discrete Fourier Transform -spread OFDM

Zero-tail Discrete Fourier Transform -spread OFDM (ZT DFT-S-OFDM) modulation allows to dynamically coping with the delay spread of the multipath channel, thus avoiding the limitations of hard-coded Cyclic Prefix (CP). The ZT DFT-S-OFDM is modulation for the envisioned 5th generation (5G) radio access technology, characterized by an ultra-dense deployment of small cells and the support of novel paradigms such as Device-to-Device (D2D). ZT DFT-S-OFDM in terms of coexistence among devices operating over adjacent frequency chunks, possibility of adopting unified radio numerology among different cells, reduced latency and support of agile link direction switching.

1.9 DFT-PRECODED OFDM

DFT-PRECODED OFDM is also called Generalized Precoded OFDMA (GPO). The waveform introduced using a frequency domain pulse shaping filter (with or without excess bandwidth) that is implemented as a baseband module. A Generalized- precoded-OFDM (GPO) waveform with a very low peak-to-average-power-ratio (PAPR) is proposed. In this method, data of multiple users are multiplexed in frequency domain using orthogonal or non-orthogonal subcarrier mapping. It comprises of: constellation rotation of user data, DFT precoding and spreading, user specific frequency domain pulse shaping (FDPS) possibly with certain excess bandwidth. The frequency domain pulse shaping filter (FDPSF) introduces inter-symbol-interference (ISI) and sometimes inter-user interference. For the special case of binary modulation, a waveform with low PAPR using frequency domain pulse shaping filters (FDPSFs) that are derived from the linearized Gaussian pulse is also proposed. To mitigate the ISI and self-interference caused by the FDPSF, conventional and widely linear frequency domain equalization methods that have low-implementation complexity is considered.