Title:  Self Calibrating Mixed Signal/RF Systems: Offline and Online Tuning Algorithms and Infrastructure

Committee: 

Dr. Chatterjee, Advisor  

Dr. Hasler, Chair

Dr. Shaolan Li

Abstract: The objective of the proposed research is to develop algorithms and infrastructure that introduce self-adaptability in modern embedded mixed signal/RF systems. With the advances in Silicon technology, the integration capabilities are increased and the transistor sizes are down to nanometer regime. On the other hand, these advances bring forth several issues such as increased parasitics, increased intra-die, intra-wafer coupling, reduced visibility to internal circuit nodes. Besides, the applications of embedded mixed signal/RF systems such as 5G and 6G wireless communication technologies that employ complex beamforming antenna arrays have stringent performance requirements under dynamically varying environmental conditions. The increased dynamic range of channel Signal to Noise Ratio (SNR) values that can be accommodated, creates power stress on Radio Frequency (RF) electronic circuitry. Further, testing and tuning of the underlying RF/mixed-signal circuitry is expensive. State of the art parallel testing and tuning techniques are expensive and limited in terms of number of components that can be tested/tuned simultaneously and in terms of the order of distortions that can be estimated. The main aim of this research is to design algorithms to develop built-in testable and self-calibrating mixed signal systems that can self-test and self-adapt on-chip with minimal use of external test instrumentation and response measurement systems. Offline parallel frequency-efficient test schemes that can detect upto fifth order non-linearities are proposed and those that can test several RF chains simultaneously in minimal time are proposed. Besides, the post manufacture tuning algorithm proposed is expedited using the information obtained by using the parallel test schemes. Real time adaptation algorithms that optimize power consumption or Energy/bit or Throughput/Watt for mixed-signal systems deployed in applications like vehicular communication and wireless communication with specifications such as bit error rate (BER) are proposed. Future work envisions addressing certain challenges in proposed algorithms and intelligent techniques for self-calibration of mixed-signal systems.