Conditions for a Monotonic Channel Capacity

Abstract

Abstract-It is proved that for memoryless vector channels, maximizing the mutual information over all source distributions with a certain average power or over the larger set of source distributions with upperbounded average power yields the same channel capacity in both cases. Hence, the channel capacity cannot decrease with increasing average transmitted power, not even for channels with severe nonlinear distortion.

Influence of Behavioral Models on Multiuser Channel Capacity

“…We know for sure that M -PSK are not optimal constellations 4 , but they suffice to show the qualitative trend of the achievable rate: It again grows with increasing power towards infinity. This result is significantly stronger than the theoretical prediction for this behavioral model with arbitrary channel models [12], which only states that the achievable rate is nondecreasing.…”

“…This result is not surprising, since X 1 dominates over the two other terms in (1) at sufficiently high power P 1 . The channel is in fact a singleuser channel, described by a fixed distribution f Y1|X1 , and the channel capacity is nondecreasing for all such channels, linear or nonlinear [12]. The capacity is larger in the case of PSfrag replacements PSfrag replacements PSfrag replacements discrete input distributions for the interfering channels than in the Gaussian case, but the capacity follows the same general trend in both cases.…”

“…However, more or less all such plots formally represent lower bounds on the channel capacity, as pointed out, e.g., in [4], [8]- [10], since they are obtained from analysis over a finite set over all possible input distributions or using suboptimal (mismatched) receivers. It is possible to show that the channel capacity will not decay at high signal powers, provided that a sufficiently exhaustive search over input distributions is carried out at each signal power level [11], [12]. Moreover, it can be shown that the use of a finite-memory channel model will also raise the lower capacity bounds at high signal powers to nonzero values [13].…”

Next-to-leading-order corrections to capacity for a nondispersive nonlinear optical fiber channel in the intermediate power region

“…Of course, the problem of capacity calculation for these channels is simpler than the problem with dispersion. However it is still quite a challenging problem especially at large parameter SNR, and new techniques and methods are highly desirable to advance these studies [3,16,17,[19][20][21].…”

Implications of information theory in optical fibre communications

“…The past 15 years have witnessed an intense research in the area of optical fibre capacity. For extensive literature surveys, see [3][4][5][6] and [7,§VI].…”