Low-power chirp signals for wireless data communication in industrial environments

A class of wideband chirp (linear FM) signals, called M-ary Chirp Spread Spectrum (MCSS) modulation, is proposed for wireless data transmission in industrial environments that have strict regulations of radiated power from electronic devices, systems, etc. such as in nuclear power plants. The MCSS modulation is described and the resulting signals are illustrated as a function of modulation parameters. The electromagnetic interference (or noise) in industrial environments is predominantly impulsive in character and is modeled as non-Gaussian statistical distribution. The dissertation consists of two parts. In the first part, optimal low-SNR receivers are derived for coherent and non-coherent detections of MCSS signals in arbitrary additive, white, non-Gaussian interference. These receivers consist of zero-memory nonlinearity (ZMNL), -ddr ln pW(r), where pW(r) is the first-order noise density, followed by corresponding coherent and non-coherent Gaussian receivers. Asymptotic bit error rate (BER) performance bounds of these receivers are derived and illustrated as functions of: (i) Signal-to-Noise Ratio (SNR), (ii) Modulation parameters; (iii) Parameters of first-order noise density; and (iv) Detection sample size. Optimal coherent MCSS modulations that minimize BER performance have been determined and illustrated. Sensitivity of optimal low-SNR receivers to uncertainties in the estimation of noise density parameters is examined, and it is found that these receivers are nearly optimal over a wide range of noise density parameters of practical interest. Also, three easy-to-implement suboptimal receivers are considered and their relative performances to that of the optimal receiver are evaluated. Finally, an investigation of error rate performance of coherent low-SNR receivers for Bose-Chaudhuri-Hocquenghem (BCH) coded MCSS modulation is carried out and performance trade-offs between coding and modulation are highlighted. Three well-known non-Gaussian noise models that represent real-world noise environments are used in this study to illustrate performance results. A comparison of error rate performance of MCSS modulation with that of conventional modulations, with and without BCH coding, is also provided.;Key words: M-ary Chirp spread spectrum, Industrial environments, Non-Gaussian noise, Locally optimum Bayes detector, Coherent and non-coherent detections, Wireless sensor networks.;In the second part of the dissertation, experimental characterization of wireless sensor networks (WSNs) is carried out in indoor and industrial environments in the presence of coexisting electromagnetic interference (EMI) on the 2.4 GHz industrial, scientific, and medical (ISM) band. The first WSN uses CSS modulation and the second uses offset-quadrature phase shift keying (O-QPSK) modulation. The performance is evaluated for a point-to-point topology with the following three scenarios: (i) when coexisted with other wireless systems, (ii) under time-varying EMI, and (iii) when the system is deployed in harsh industrial environment. In these evaluations, the transmission power levels are very low in the range from 0 dBm to -28.9 dBm. The performance metrics used are: (i) packet loss rate; (ii) packet error rate; (iii) bit error rate; and (iv) throughput. The results obtained are particularly applicable in the development of low-power and high-reliability wireless network systems in industrial environments in which communication systems are required to meet strict regulations in terms of radiated power.