Abstract
The exploration by expert systems of the information present in the signal constellation of a 16-QAM radio for fault diagnosis is investigated. A 16-QAM radio model was developed on a nonintrusive communications analyzer for this purpose. Two approaches are under investigation: a conventional rule-based system and a machine-learning system based on adaptive pattern-recognition techniques. Both techniques are described, and their relative performance is compared using data from the radio model. Typical synthesized faults are up to 8-dB TWT overdrive and underdrive, up to 5 degree nonorthogonality of I and Q carriers, and up to plus or minus 10% spacing error in the constellation. Further information is provided on the machine learning system's performance on a real radio operating at 11 GHz when feeding back to itself at RF.
| Original language | English |
|---|---|
| Pages (from-to) | 1159-1163 |
| Number of pages | 5 |
| Journal | IEEE International Conference on Communications |
| State | Published - 1987 |
ASJC Scopus subject areas
- Computer Networks and Communications
- Electrical and Electronic Engineering