Robust Non-Data-Aided SNR Estimation for Multilevel Constellations via Kolmogorov–Smirnov Test

Reduced Complexity SNR Estimation via Kolmogorov-Smirnov Test

“…Assuming a quasi-static flat-fading channel model, the symbol rate samples at matched filter output are given by [7], i.e.,…”

“…The K-S test based SNR estimation procedure proposed in [7] can be described as follows briefly: 1) Upon the constellation format, choose a proper decision statistic z (actually, the signal envelope is adopted) and compute its theoretical…”

“…As has been stated in [7], the K-S test based SNR estimator involves no multiplications but considerably more real additions (K log 2 K + 2KK snr − 1) and memory (K snr N ref ) are needed because of the large amount of reference signals and comparators. Usually, the K snr should be large enough for accurate estimates and N ref should be large for a smooth CDF.…”

“…But the weights of the linear combination should be tuned according to the constellation and the specific SNR operation range. In [7], we have proposed a novel Kolmogorov-Smirnov (K-S) test based NDA SNR estimator, which can work properly over an extended SNR range for various multilevel constellations with limited signal samples. As to the computation complexity, the K-S test based estimator involves no multiplication operations, but considerably more add operations are needed, which requires a large amount of logic resources in hardware.…”

Shrinkage‐based biased signal‐to‐noise ratio estimator using pilot and data symbols for linearly modulated signals

“…The conditional expectation in E-step can then be represented in the same manner as in the previous section (see (10)) where M 2,d = 1/L L k=1 |z k | 2 and M 2,p = 1/P L+P k=L+1 |z k | 2 . Differentiating (10) with respect to h and s 2 n leads to the following resultŝ…”