Open AccessErgodic sum rate analysis and efficient power allocation for a massive MIMO two‐way relay network
Author(s) -
Roshandeh Koosha Pourtahmasi,
Kuhestani Ali,
Ardakani Masoud,
Tellambura Chintha
Publication year - 2017
Publication title -
iet communications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.355
H-Index - 62
eISSN - 1751-8636
pISSN - 1751-8628
DOI - 10.1049/iet-com.2015.1029
Subject(s) - relay , beamforming , ergodic theory , mimo , computer science , nakagami distribution , channel (broadcasting) , transmission (telecommunications) , transmitter power output , mathematical optimization , maximal ratio combining , signal to noise ratio (imaging) , closed form expression , power (physics) , topology (electrical circuits) , mathematics , telecommunications , fading , combinatorics , transmitter , mathematical analysis , physics , quantum mechanics
The authors study the transmit power allocation (PA) problem for a network of two multi‐antenna terminals (one of which is a massive multiple‐input and multiple‐output (MIMO) terminal) and a two‐way, amplify‐and‐forward relay. The relay is limited to a single antenna. Using perfect channel state information, the terminals employ beamforming with maximum‐ratio‐transmission and maximum‐ratio‐combining for transmission and reception, respectively. The authors investigate two practical problems, namely; (i) maximising the sum rate subject to a total power constraint (ii) maximising the sum rate when one of the terminals must exceed a target signal‐to‐noise ratio (SNR). For the first case, the authors derive the closed‐form optimal PA and for the second, the authors derive a sub‐optimal PA. In both cases, the resulting sum rates are a function of instantaneous channel gains. Thus by averaging over the Nakagami‐ m distribution and exploiting the weak law of large numbers, the authors derive the closed‐form ergodic sum rates. Finally, the simulation results validate the theoretical analysis and show the sum‐rate improvements over uniform PA. For example, to achieve 4 bit/s/Hz, a uniform allocation needs 1 dB more than the authors’ optimal allocation. When one of the SNRs must exceed a target value, the gap between the authors’ sub‐optimal PA and random PA increases to 2 dB.