Exploiting Spatial Degrees-of-Freedom for Energy-Efficient Next Generation Cellular Systems

This research addresses green communication issues, including energy efficiency, peak-to-average power ratio (PAPR) reduction and power amplifier (PA) linearization. Green communication is expected to be a primary goal in next generation cellular systems because it promises to reduce operating costs.

The first key issue is energy efficiency of distributed antenna systems (DASs). The power consumption of high power amplifiers (HPAs) used in wireless communication systems is determined by the transmit power and drain efficiency. For unequal power allocation of orthogonal frequency division multiplexing (OFDM), the drain efficiency of the PA is determined by the PAPR and hence by the power distribution. This research proposes a PAPR-aware energy-efficient resource allocation scheme for joint orthogonal frequency division multiple access (OFDMA)/space division multiple access (SDMA) downlink transmission from DASs. Grouping-based SDMA is applied to exploit the spatial diversity while avoiding performance degradation from correlated channels. The developed scheme considers the impact of both system data rate and effective power consumption on the PAPR during resource allocation. We also present a suboptimal joint subcarrier and power allocation algorithm to facilitate implementation of power-efficient multi-channel wireless communications. By solving Karush-Kuhn-Tucker conditions, a closed-form solution for the power allocation of each remote radio head is obtained.

The second key issue is related with PAPR reduction in the massive multiple-input multiple-output (MIMO) systems. The large number of PAs in next generation massive MIMO cellular communication system requires using inexpensive PAs at the base station to keep array cost reasonable. Large-scale multiuser (MU) MIMO systems can provide extra spatial degrees-of-freedom (DoFs) for PAPR reduction. This work applies both recurrent neural network (RNN)- and semidefinite relaxation (SDR)-based schemes for different purposes to reduce PAPR. The highly parallel structure of RNN is proposed in this work to address the issues of scalability and stringent requirements on computational times in PAPR-aware precoding problem. An SDR-based framework is proposed to reduce PAPR that accommodates channel uncertainties and intercell coordination. Both of the proposed structures reduce linearity requirements and enable the use of lower cost RF components for large-scale MU-MIMO-OFDM downlink.

The third key issue is digital predistortion (DPD) in the massive MIMO systems. The primary source of nonlinear distortion in wireless transmitters is the PA, which is commonly modeled using polynomials. Conventional DPD schemes use high-order polynomials to accurately approximate and compensate for the nonlinearity of the PA. This is impractical for scaling to tens or hundreds of PAs in massive MIMO systems. This work therefore proposes a scalable DPD method, achieved by exploiting massive DoFs of next generation front ends. We propose a novel indirect learning structure which adapts the channel and PA distortion iteratively by cascading adaptive zero-forcing precoding and DPD. Experimental results show that over 70% of computational complexity is saved for the proposed solution, it is shown that a 3rd order polynomial with the new solution achieves the same performance as the conventional DPD using 11th order polynomial for a 100x10 massive MIMO configuration.