Infinite-SNR asymptotic MMSE limit

Let D(R,β)D(R,\beta) denote the asymptotic MMSE at measurement rate RR and inverse noise variance (SNR) parameter β\beta, and let mam_a be the a-priori magnetization. In the low-noise regime, consider β\beta\to\infty. Infinite SNR limit. For R>maR>m_a, the asymptotic MMSE satisfies

limβ[βD(R,β)]=σ2maRma.\lim_{\beta\to\infty}\left[\beta\cdot D(R,\beta)\right]=\sigma^2\frac{m_a}{R-m_a}.

This conjecture describes the noise sensitivity in the high-SNR regime when the measurement rate exceeds the a-priori magnetization. The preceding discussion indicates that the asymptotic MMSE is governed by the first term in the MMSE expression in this range; the corresponding behavior for R<maR<m_a is that βD(R,β)\beta\cdot D(R,\beta) grows without bound, consistent with earlier observations for independent and identically distributed sources.

Sources & referencesView supporting material

Primary source

Wasim Huleihel and Neri Merhav, “Asymptotic MMSE Analysis Under Sparse Representation Modeling”, arXiv:1312.3417 (2016).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.