Analysis of Alternative Metrics for the PAPR Problem in OFDM Transmission

TL;DR

This paper investigates alternative metrics for PAPR in OFDM systems, revealing that the traditional log(N) barrier is overly conservative and proposing new bounds and relationships for these metrics.

Contribution

The paper provides the first theoretical analysis of alternative PAPR metrics, showing they scale more slowly than the traditional log(N) barrier.

Findings

Log(N) barrier is overly conservative for PAPR scaling.

Amplifier-oriented metric scales as log(log(N)).

New upper bounds on PAPR distribution are established.

Abstract

The effective PAPR of the transmit signal is the standard metric to capture the effect of nonlinear distortion in OFDM transmission. A common rule of thumb is the log$(N)$ barrier where $N$ is the number of subcarriers which has been theoretically analyzed by many authors. Recently, new alternative metrics have been proposed in practice leading potentially to different system design rules which are theoretically analyzed in this paper. One of the main findings is that, most surprisingly, the log$(N)$ barrier turns out to be much too conservative: e.g. for the so-called amplifier-oriented metric the scaling is rather $\log[ \log(N)]$. To prove this result, new upper bounds on the PAPR distribution for coded systems are presented as well as a theorem relating PAPR results to these alternative metrics.