An Overview of Low-Rank Channel Estimation for Massive MIMO Systems
Citations
577 citations
Cites background from "An Overview of Low-Rank Channel Est..."
...However, massive MIMO encounters some practical challenges, including the sophisticated channel modeling, the high-dimensional channel state information (CSI), the scheduling of numerous accessing users and the limited radio frequency (RF) chains, etc [10]....
[...]
493 citations
250 citations
Cites background from "An Overview of Low-Rank Channel Est..."
...However, when M and N are finite in practice, the region of the non-zero square will be expanded due to the power leakage effect [15], [31]....
[...]
...Note that the proposed soft grouping strategy is different from the conventional hard grouping one [11], [15], [31] in that the latter requires users with overlapped signature to transmit over completely non-overlapped time intervals while ours allows users to transmit in an “interlock” way by purposely postpone certain users such that (29) can be satisfied....
[...]
...The extension to low frequency-band communications can be straightforwardly made by counting the angular spread caused by local scattering [15], [31]....
[...]
197 citations
Cites background from "An Overview of Low-Rank Channel Est..."
...The massive MIMO regime enables tens of hundreds/thousands antennas at a BS, and hence it is capable of offering an unprecedented level of freedom to serve multiple mobile users [14]....
[...]
183 citations
References
12,446 citations
6,273 citations
"An Overview of Low-Rank Channel Est..." refers background in this paper
...However, there are three main reasons that the conventional MUSIC [12] and ESPRIT [13] are not applicable here: (i) They may suffer from very high computational complexity due to their SVD operation with massive antennas; (ii) They are designed for the scenario when the incoming signals do not have AS and would suffer from performance degradation with surrounding scattering; (ii) They are blind approaches originally designed for Radar application but do not utilize the training sequences embedded in communications systems....
[...]
...ANTENNA ARRAY THEORY BASED LOW-RANK CHANNEL ESTIMATION Now that the angular information is crucial for the low-rank channel estimation, a natural question arise: why don’t we directly achieve such angular information via certain canonical means, say, array signal processing? However, there are three main reasons that the conventional MUSIC [12] and ESPRIT [13] are not applicable here: (i) They may suffer from very high computational complexity due to their SVD operation with massive antennas; (ii) They are designed for the scenario when the incoming signals do not have AS and would suffer from performance degradation with surrounding scattering; (ii) They are blind approaches originally designed for Radar application but do not utilize the training sequences embedded in communications systems....
[...]
6,248 citations
"An Overview of Low-Rank Channel Est..." refers background in this paper
...If the same training sequences are reused or non-orthogonal training sequences are adopted, then the inter-user interference will arise during the channel estimation stage, which is known as pilot contamination [1]....
[...]
...Unfortunately, channel reciprocity is not applicable for frequency division duplexing (FDD) systems, which is still a dominant transmission mode in most communications systems [1]....
[...]
...INTRODUCTION Large-scale multiple-input multiple-output (MIMO) or ‘‘massive MIMO’’, a new technique that employs hundreds or even thousands of antennas at base station (BS) to simultaneously serve multiple users, has been widely investigated for its numerous merits, such as high spectrum and energy efficiency, high spatial resolution, and simple transceiver design [1]....
[...]
4,122 citations
2,466 citations
"An Overview of Low-Rank Channel Est..." refers background in this paper
...From the conventional orthogonal training strategy [2], the required number of orthogonal training sequences as well as the length of the training sequences should be at least the number of transmit antennas....
[...]