READ ME File For 'Multicarrier Division Duplex Aided Millimeter Wave Communications'

IEEE Access (Accepted on 18 July 2019)

Authors: R. Rajashekar, C. Xu, N. Ishikawa, L-L. Yang and L. Hanzo

R. Rajashekar, C. Xu,  L-L. Yang and L. Hanzo are with the School of Electronics and Computer Science, University of Southampton, Southampton SO17 1BJ, UK (e-mail: \{rmr1u14,cx1g08,lly,lh\}@soton.ac.uk). N. Ishikawa is with the Graduate School of Information Sciences, Hiroshima City University, Ohzuka-higashi 731-3194, Japan (e-mail: naoki@ishikawa.cc).

Acknowledgement: 
This work was supported in part by the EPSRC projects EP/Noo4558/1 and EP/PO34284/1, the European Research Council's Advanced Fellow Grant under the QuantCom project and the Royal Society's Wolfson Research Merit Award and the Royasl Society's Global Research Challenges Fund . The work of N. Ishikawa was supported in part by the Japan Society for the Promotion of Science KAKENHI under Grant 17H07036.

Abstract:
The existing Time Division Duplex (TDD) and Frequency Division Duplex (FDD) techniques rely on a guard time and/or guard band to avoid Self-Interference (SI) between the uplink and downlink channels, which results in the wastage of precious spectral resources. The Full-Duplex (FD) schemes of In-Band Full Duplex (IBFD) as well as Multicarrier Division Duplex (MDD) may overcome this drawback while retaining the key benefits of both TDD and FDD. Moreover, the MDD exhibits the exclusive benefits of the reduced Peak-to-Average Power Ratio (PAPR) for signal transmission as well as the SI-free signal detection. Against this background, in this work, we propose a novel FD scheme conceived for frequency-selective millimeter Wave (mmWave) channels, which have not been investigated in the open literature. Furthermore, we propose a novel Projection aided Iterative Eigenvalue Decomposition (P-IEVD) algorithm, which performs nullspace SI cancellation in the inherent beamforming structure of mmWave communication. Our simulation results confirm that the MDD is capable of outperforming its Half-Duplex (HD) counterparts of TDD/FDD, even the IBFD can only achieve a better bandwidth efficiency than the MDD when a sufficiently high SNR is provided.

Fig.~1:
Schematic_TDD_FDD_IBFD_MDD.eps

Fig.~2:
System_Model.eps

Fig.~3:
Tx_Rx_ULAs.eps

Fig.~4:
PA_Response.eps

Fig.~5:
PAPR_IBFD_MDD_QPSK.eps
PAPR_IBFD_MDD_N_64.eps

Fig.~6:
ADC_Dynamic_Range.eps

Fig.~7:
SQNR_FD_Q_d_5.eps
SQNR_FD_Q_8_d.eps

Fig.~8:
Schematic_HD_FD_Beamforming.eps

Fig.~9:
BF_Gain_HD_Nt_8_Nr_4_Nc_5_Nray_8_Dmax_8_NOFDM_64_rhoii_15_rhoiiLoS_50.eps
BF_Gain_MDD_FD_HD_Nt_8_Nr_4_Nc_5_Nray_8_Dmax_8_NOFDM_64_rhoii_15_rhoiiLoS_50.eps

Fig.~10:
SI_LoS_MDD_FD_HD_Nt_8_Nr_4_Nc_5_Nray_8_Dmax_8_NOFDM_64_rhoii_15_rhoiiLoS_50.eps

Fig.~11:
CCMC_Capacity_HD_Training_Nt_8_Nr_4_SI_15_IEVD_Training.eps
CCMC_Capacity_MDD_Training_Nt_8_Nr_4_SI_15_IEVD_Training.eps

Fig.~12:
CCMC_Capacity_MDD_Nt1_8_Nt2_4_Nr_2.eps
CCMC_Capacity_MDD_Nt_8_Nr1_6_Nr2_4.eps

Fig.~13:
CCMC_Capacity_MDD_FD_Nt_8_Nr_4_IEVD_itera_2.eps
CCMC_Capacity_MDD_HD_Nt_8_Nr_4_IEVD_itera_2.eps


