OAI based Universal Filtered Multi-Carrier (UFMC) System
Authors: Carmine Vitiello, Raymond Knopp, Florian Kaltenberger
5G promises the biggest revolution in the world of communication systems, where many new technologies and services will be included into the next communication standard. In particular, Internet of Things (IoT) will bring these technologies at any time and any place in our lives, starting from the daily routine, such as the domestic appliances or home automation, to more futuristic application such as tele-medicine, secure vehicular communication and so on. Given these huge changes, the whole stack of the mobile communication network must be re-thought from scratch. In terms of physical layer, the current waveform, Orthogonal Frequency Division Multiplexing (OFDM), shows some weak aspects, which appears unsuitable to support new requirements. One of the main requirements is given by the minimization of end-to-end latency up to millisecond, which will allow the diffusion of tactile internet. A fast access to the wireless medium is also necessary for avoiding useless idle times in the synchronization procedure, allowing a fast dormancy and an asynchronous and non-orthogonal access of the medium . Furthermore, the new waveform would better exploit the blank spaces or the space between adjacent channel, for saving the precious spectrum resource and finally allowing an opportunistic access, enabling the cognitive radio communications. It is clear not every services or applications needs the same requirements; therefore the selected waveform would support different quality of services, as requested by heterogeneous nature of 5G. Moreover, the requirements would be satisfied keeping in mind the complexity of the operation at transmitter and receiver, especially with the exponential growth of the entities that potentially could be involved in a communication and preserving the backward compatibility with previous generation of wireless mobile network. In this regard, many new waveforms have been proposed as new candidate to be the new physical layer of 5G, each one with several strong and weak aspects. However, in this project we implemented Universal Filtered MultiCarrier (UFMC)  within OpenAirInterface .
In contrast to OFDM, UFMC groups sub-carriers into several sub-bands, applying filtering on them separately. In this way, it is possible to achieve robustness against time and frequency misalignment without any overhead in cyclic prefix and so on, improving spectral efficiency and reducing out-of-band emission without losing OFDM advantages. All of these qualities are particularly applicable in a typical IoT scenario, where small amount of data is sporadically generated by an IoT device. First of all, the waveform guarantees a well-defined spectrum occupation without interference with adjacent communications and therefore avoiding any waste of energy, as depicted in Fig.1. In addition, our work has been focused on improving the computational complexity for carrying few symbols, achieving better performance than both original implementation and OFDM, as depicted in Fig 2. This result has been obtained thanks to a reduction of the IFFT order, from N to only Nr, and the correct usage of upsampling. Furthermore, a real-coefficient version of a Chebyshev filter with length L works in time domain only on non-zero samples, followed by a shifting operation for moving the signal in correct sub-band. In this way, large amount of the processing has been performed in baseband, while at the same time reducing the modulator operations. In order to prove its quality in real world and for assessing the coexistence with the OFDM, UFMC waveform has been implemented on an open-source implementation of LTE 4G system called OpenAirInterface . This framework has allowed easy and fast implementation, while at the same time providing the whole stack of the 3GPP 4G/5G cellular communication standard. Specifically, it has been possible to replace the SC-FDMA waveform with a single-carrier version of UFMC, named SC-UFMC, in the uplink communications of the ULSCH/PUSCH channel with 10MHz bandwidth . The waveform coexistence has been proved just changing the modulator at the transmitter side, with the new one in Fig.3(b). In the UFMC modulator, each subband takes the same meaning of a Physical Resource Block (PRB), therefore they group 12 contiguous sub-carriers. As already mentioned, IFFT size moves from N = 1024 to only Nr = 64, allowing a notable gain, while filter length has been fixed to the cyclic prefix length of LTE system minus one, therefore 73 or padded up to 80 in case of extended prefix in order to get same output length of OFDM transmitter, yielding transparent change in the modulator. At the receiver, we have maintained same structure of LTE compliant PUSCH receiver, composed by a 1024-FFT followed by an equalizer, and including a timing synchronization block, which is able to estimate the delay correctly for performing FFT operation. Timing synchronization is achieved by exploiting uplink demodulation reference signal (DRS), sent out as third multicarrier symbol of each slot of an LTE subframe. Assuming that the receiver doesn’t know the delay of the channel, it takes complex symbols of received vector within a window with dimension 2N around the position of the third multicarrier symbol of each slot calculated in absence of channel delay. So a 2N-FFT is performed, then multiplied by the conjugate DRS sequence already processed in the same way as UFMC signal. Absolute square is performed followed by an element-wise sum of the vectors. At the end, a peak detection reveals the exact delay from which is possible to apply regular PUSCH receiver. Receiver structure is represented in fig.4. This kind of timing synchronization, joined with the nature of the waveform, improves the robustness against time mis-alignments at low SNR. This feature is exploitable for modifications to the synchronization procedure, leaving the classical RACH procedure for an asynchronous transmissions, more suitable for IoT devices such as contention-based access procedure  and already implemented in OpenAirInterface. Furthermore, the described receiver is not the optimal one but it is an ad-hoc solution just for maintaining the coexistence between the two waveforms at the cost of small degradation in performance. Despite this adaptation, UFMC still shows performance quite similar to OFDM.
UFMC promises to be one of the best candidate technologies for supporting the physical layer of next-generation wireless mobile network. In particular, its features are quite suitable as the main driver of IoT application scenarios. Furthermore, its implementation within OpenAirInterface has shown to bring this new improvement also in 4G, proving the coexistence with the existing waveform and represents one of the best examples of new features integration within the current standard. We would like to extend this work with the community towards implementing 3GPP compliant new waveforms within OpenAirInterafce  once they are standardized in future releases.
 Wunder, G. et al, ”5GNOW: non-orthogonal, asynchronous waveforms for future mobile applications”, in Comm. Magazine, IEEE, Feb. 2014.
 Wang, X.; Wild, T.; Schaich, F. and Fonseca dos Santos, A., ”Universal Filtered Multi- Carrier with Leakage-Based Filter Optimization”, in European Wireless Conference 2014; Proceedings of, May 2014.
 Knopp, R.; Kaltenberger, F; Vitiello, C and Luise, M.,”Universal filtered multicarrier for machine type communications in 5G”, EuCNC 2016; Proceedings of, June 2016.
 LTE E-UTRA; Requirements for support of radio resource management (3GPP TS 36.133 version 12.10.0 Release 12).
 K. Zhou, N. Nikaein, R. Knopp and C. Bonnet, ”Contention Based Access for Machine- Type Communications over LTE”, Vehicular Technology Conference (VTC Spring), 2012 IEEE 75th, 2012.
 OpenAirInterface Software Alliance, https://openairinterface.org