Abstract
The downlink (DL) of a non-orthogonal-multiple access (NOMA)-based cell-free massive multiple-input multiple-output (MIMO) system is analyzed, where the channel state information (CSI) is estimated using pilots. It is assumed that the
users are grouped into multiple clusters. The same pilot sequences are assigned to the users within the same clusters whereas the pilots allocated to all clusters are mutually orthogonal. First, a user’s bandwidth efficiency (BE) is derived based on his/her channel statistics under the assumption of employing successive
interference cancellation (SIC) at the users’ end with no DL training. Next, the classic max-min optimization framework is invoked for maximizing the minimum BE of a user under per access point (AP) power constraints. The max-min user BE of NOMA-based cell-free massive MIMO is compared to that of
its orthogonal multiple-access (OMA) counter part, where all users employ orthogonal pilots. Finally, our numerical results are presented and an operating mode switching scheme is proposed based on the average per-user BE of the system, where the mode set is given by Mode = { OMA, NOMA }. Our numerical results confirm that the switching point between the NOMA and OMA modes depends both on the length of the channel’s coherence time and on the total number of users.
users are grouped into multiple clusters. The same pilot sequences are assigned to the users within the same clusters whereas the pilots allocated to all clusters are mutually orthogonal. First, a user’s bandwidth efficiency (BE) is derived based on his/her channel statistics under the assumption of employing successive
interference cancellation (SIC) at the users’ end with no DL training. Next, the classic max-min optimization framework is invoked for maximizing the minimum BE of a user under per access point (AP) power constraints. The max-min user BE of NOMA-based cell-free massive MIMO is compared to that of
its orthogonal multiple-access (OMA) counter part, where all users employ orthogonal pilots. Finally, our numerical results are presented and an operating mode switching scheme is proposed based on the average per-user BE of the system, where the mode set is given by Mode = { OMA, NOMA }. Our numerical results confirm that the switching point between the NOMA and OMA modes depends both on the length of the channel’s coherence time and on the total number of users.
Original language | English |
---|---|
Pages (from-to) | 792 - 810 |
Journal | IEEE Transactions on Communications |
Volume | 68 |
Issue number | 2 |
Early online date | 11 Nov 2019 |
DOIs | |
Publication status | Published - Feb 2020 |