Zipf's law and tactical/mobile speech communications
D. Yavuz · 2002
Zipf (1949) has shown that if the words in a language are ordered (ranked) in terms of frequency of usage, the relationship between probability of use and rank is an inverse power law with unit exponent. Shannon (1951) has used this to estimate the entropy of English but with some numerical errors which are corrected. The implications of these results to mobile/tactical communications over, non-line-of-sight, limited capacity radio channels is presented. Such very low rate speech coding will not provide speech waveform fidelity but will provide content fidelity which can be acceptable is some scenarios. The advantages gained in terms of jamming margins and performance over difficult non-line-of-sight are substantial. It is concluded that there is still some potential for speech coding bit rate reduction for use in such difficult channels.