Abstract: For a multiple antenna system, we compute the asymptotic distribution of antenna selection gain when the transmitter selects the transmit antenna with the strongest channel. We use this to asymptotically estimate the underlying channel capacity distributions, and demonstrate that unlike multiple-input/multiple-output (MIMO) systems, the channel for antenna selection systems hardens at a slower rate, and thus a significant multiuser scheduling gain can exist - O(1/ log m) for channel selection as opposed to O(1/ sqrt{m}) for MIMO, where m is the number of transmit antennas. Additionally, even without this scheduling gain, it is demonstrated that transmit antenna selection systems outperform open loop MIMO systems in low signal-to-interference-plus-noise ratio (SINR) regimes, particularly for a small number of receive antennas. This may have some implications on wireless system design, because most of the users in modern wireless systems have low SINRs.
Introduction
In Multiple-Input Multiple-Output (MIMO) wireless systems, the addition of more antennas typically leads to a phenomenon known as "channel hardening." As the number of antennas increases, the random fluctuations of the fading channel tend to average out, causing the channel to behave more like a deterministic, non-fading one. While this predictability can simplify some aspects of system design, it significantly reduces "multiuser scheduling gain"—the ability of a base station to exploit peaks in a specific user's channel by scheduling transmissions during those moments of high gain.
This paper investigates whether transmit antenna selection (TAS), where the transmitter chooses only the single best antenna based on channel state information, suffers from the same rapid hardening. Furthermore, it explores whether this lower-complexity alternative can compete with or even outperform full-blown MIMO systems, particularly in the low Signal-to-Interference-plus-Noise Ratio (SINR) regimes common in modern mobile networks.
Analysis of Selection Gain
The researchers utilize extreme order statistics to analyze an $n \times m$ MIMO channel model. In this setup, the transmitter selects the single antenna that provides the strongest total channel gain. A key theoretical contribution is the proof that the distribution of this selection gain converges to a Gumbel distribution (a type of Generalized Extreme Value distribution) as the number of transmit antennas $m$ increases.
To ensure the model's utility in practical scenarios with a finite number of antennas, the authors derived optimal normalizing constants ($a_m$ and $b_m$). These constants provide a remarkably accurate fit for the Gumbel distribution even when $m$ is small.
As shown in Figure 1, the Gumbel approximation closely tracks the exact cumulative distribution function (CDF) of the selection gain across different numbers of receive antennas ($n$) and transmit antennas ($m$).
Channel Hardening and Scheduling
A critical insight of this work is the difference in hardening rates. While standard MIMO channels harden at a rate of $O(1/\sqrt{m})$, antenna selection systems harden much more slowly, at a rate of $O(1/\log m)$. This implies that the "peaks" and "valleys" of the channel gain persist even as more antennas are added.
Because the channel does not flatten out as quickly, a greedy scheduler can find significantly higher peaks among multiple users. The paper quantifies this, showing that the fractional scheduling gain for antenna selection is $O(1/(\log m \log(\log m)))$, whereas for standard MIMO, this gain asymptotically vanishes.
Capacity and Performance Comparison
Beyond scheduling advantages, the paper demonstrates that TAS can be superior in low SINR environments. The authors provide tight bounds for ergodic capacity, showing that the capacity of antenna selection continues to grow with $m$, unlike open-loop MIMO (where the transmitter lacks channel information), which is fundamentally limited by the number of receive antennas.
Figure 2 illustrates the 10% outage capacity as a function of the number of transmit antennas. The results show that TAS (both exact and approximated) consistently outperforms or matches simulated MIMO/MISO performance at an SINR of 5dB, especially as the number of transmit antennas increases. This is particularly relevant for modern wireless systems where a majority of users operate under low SINR conditions.
Conclusion
This research highlights that the slow channel hardening of antenna selection systems is a beneficial feature that preserves multiuser scheduling gains. By leveraging extreme value theory, the authors provide a robust framework for estimating performance. While the current analysis assumes i.i.d. Rayleigh fading and perfect feedback for a single selected antenna, it sets the stage for future investigations into subset selection and more complex channel environments.
Summary updated