Transmission energy allocation with low peak-to-average ratio

Transmission energy allocation (TEA) to bits according to their sensitivity is known to significantly enhance robustness to channel errors. These advantages are gained at the cost of high peak-to-average ratio (PAR) of the signal energy employed to transmit different bits. We show that, in the case of 4-QAM, appropriate grouping of bits allows achieving all the gains of optimal TEA while maintaining PAR at a small fraction of a decibel. Alternatively, we show how to achieve close to optimal TEA under the constraint of perfect (0 dB) PAR, thus extending the application of TEA to constant envelope modulation schemes. Performance is illustrated with an example of Gauss-Markov sources compressed by vector quantization.