Improved fuzzy frequency hopping

Abstract

Spread spectrum is a key technology in secure, wireless communications. One type of spread spectrum is frequency hopping. Each spread spectrum system uses a pseudo-random number (PN) generator to produce an output sequence which is uniformly and nonrepeatable. The most common type of PN generator is a shift register with linear feedback (LSFR). A paper by Pacini and Kosko (see IEEE Trans. Comm., vol.43, no.6, pp. 2111-17, 1995), introduced a PN generator based on fuzzy logic. Using fuzzy logic to create a PN generator has many strengths. A fuzzy system is inherently non-linear. The output sequence would be of unknown length (if it ever repeats). The fuzzy system output would be more uniform or more randomly spread. The purpose of this paper is to improve upon the Pacini and Kosko system to give a more uniform random output sequence. The system of Pcaini is limited by its initial setup parameters. An unsupervised adaptive learning algorithm, called the mountain clustering algorithm, can be used to optimize the fuzzy system (by generating a rule base and membership functions). During simulation, this approach resulted in a more uniform spread in the output sequence than an LSFR or the Pacini and Kosko system.