Signal Intelligence Demodulation Capability, 16-9514Printer Friendly Version
Inclusive Dates: 10/01/04 02/10/05
Background - Wireless communication in the high frequency (HF) portion of the spectrum presents unique challenges. Reflections from the turbulent ionosphere, a propagation mechanism not available in any other portion of the spectrum, introduce a rapidly time-varying distortion at the receiving end of a transmission. A received signal is further distorted by co-channel interference and high background noise level. This interference is caused by the crowding of many users into the relatively narrow HF range. All these factors make demodulation (extracting the information content) of a received signal difficult.
Approach - The purpose of this project was to develop an adaptive, trained equalizer to enable reliable demodulation of high baud rate HF-propagated phase shift keyed (PSK) signals. Adaptive equalizers are capable of estimating and removing the time-varying distortion mentioned above. Trained equalization refers to adaptation based on receiver knowledge of a small portion of the transmitted signal (training sequence). Work began with a bench-measured wireless signal with known content. The previously existing demodulator, which lacked an equalizer, was able to extract all the information from this clean signal without error. When an HF channel simulator conforming to international standards was used to add distortion, the existing demodulator fell apart under conditions defined as good. Trained equalization requires time synchronization to the training sequence. To achieve the necessary degree of synchronization, a new correlative method was developed. The first attempt at equalization was a conventional least mean square (LMS) algorithm. To improve robustness and decrease convergence time, a second-order LMS was implemented, and decision feedback was added.
Accomplishments - An adaptive, trained equalizer was successfully implemented within a C++ based demodulation framework developed by the Signal Exploitation and Geolocation Division. With client funding, the equalizer was subsequently incorporated into fielded systems. SwRI is currently building on this project to develop blind equalizers and demodulation capability of more complex modulation types. The correlative method for synchronizing to the training sequence has become an independent component of Sigrec, a blind signal recognition product developed by SwRI. It has been incorporated into multiple fielded systems and has increased the rate of successful recognition for traditionally difficult targets.