Nowadays, various emerging applications, services, and user requirements require rethinking the future multiple access technique of 6G. First, 6G should fully enable high mobility, using high frequency bands with large bandwidths, thanks to a strong resilience to high Doppler spreads. Second, 6G should offer massive IoT-enabled applications, with reduced frequency synchronization and power control signaling overhead, thanks to a strong resilience to out-of-band emissions, time and frequency synchronization errors and received powers imbalance.
While traditional CP-OFDM, on which 4G and 5G are based, is still suitable for many future 6G use cases, it will suffer from severe self and multi-user interference, in such high Doppler spread and out-of-band emissions use cases. Therein lies the need for technological advancements, in terms of multiple access and, more specifically, on waveform design, to unlock the true value of 6G connectivity.
The need for a strong resilience to self and multi-user interference brought several proposals for new multiple access techniques, over the last decade. Among the recently published proposals are Orthogonal Time-Frequency Space (OTFS), Sparse Code Multiple Access (SCMA), and Non-Orthogonal Waveform (NOW). Moreover, given the ascending role to be played by Machine Learning (ML) in 6G, a proposal considered learning waveform design, with ML not only replacing the processing blocks, but also the physical layer and underlying waveform and modulation.
In this talk, will be concerned with the design of new waveforms, operating on the same time-frequency layout of 4G and 5G, as potential substitutes of CP-OFDM rectangular waveforms. While offering strong resilience to self and multiple-access interference, these waveforms have the merits of easily fusing with 4G and 5G technologies and enabling 6G to be rolled out faster.
We will start by a historical overview of the Ping-pong Optimized Pulse Shaping (POPS) algorithm, and its efficient application to different waveform optimization problems, within the single-access framework, whereby a single user makes use of the whole available bandwidth. We will examine the effectiveness of several time-frequency layouts (rectangular and hexagonal) and modulations (QAM and OQAM), in the light of best achieved Signal-to-Noise Ratio (SINR), through POPS waveform optimization. We will show that the additional gains in performance, in terms of SINR, brought hexagonal lattices and OQAM are sufficiently low and do not justify the complexity they entail.
We then move to waveform design in a multiple-access context, whereby different communications share the available frequency bandwidth. We will consider several radio interface impairments, such as time-frequency synchronization errors and received power imbalance, at the receiver, in both uplink and downlink, and time asynchronism at the transmitters, in uplink. We will introduce the notion of equivalent SINR, use the POPS algorithm for waveform optimization and the CP-OFDM as benchmark. We will show the effectiveness of the designed waveforms, with respect to CP-OFDM, in terms of resilience to self and multiple access interference, especially in the uplink, when full asynchronism is assumed.