Dept. of Biomedical Engineering
Oregon Health & Science University
Analog signal representation will remain essential wherever there is a need to interface with the analog world or to satisfy certain area, power consumption, or speed requirements. This includes but is not limited to sensors, instrumentation, and communications. Analog representation is also essential for the integration of analog mixed-signal and RF functions into complex system-on-a-chip (SOC) designs. Today, analog signals are still being represented mainly by current or voltage. Also, data is usually obtained from sensors in voltage or current form. These analog signals are not immune to noise, and therefore they have to be converted into digital for transmission. In this thesis, we propose a third approach that converts the analog signal into a pulse stream, using time rather than magnitude. This alternative approach uses the inter-pulse time (IPI) to represent the signal values. The thesis will show that our representation approach, unlike the other pulse time representation approaches, is very useful not only in communication but in computation as well. Suitability for both communication and computation is very important because it eliminates the need to convert to/from the analog or digital domains to use their computation techniques if computation is needed. One good example where computation would be needed with communication is the use of averaging at the front end of the receiver to improve the signal-to-noise ratio (SNR). The thesis will also show that our approach is a hybrid approach that takes from digital the immunity to noise, cross-talk, and other problems such as process variations, temperature, and reference voltage, and takes from analog the compactness and low power consumption. In this thesis, we also present a novel class of methods and circuits for basic conversion and computation based on our novel IPI representation approach above. These methods and circuits include Voltage-to-IPI, IPI-to-voltage, addition, subtraction, division, and multiplication. We validate these methods and circuits by mathematical derivation, simulation, and chip fabrication and test in CMOS technology. We also compare our IPI implementations versus analog and digital implementations, show their advantages, and discuss how they can be used in applications such as communications, instrumentation, telemetry, signal processing, and ANNâs.
OGI School of Science and Engineering
Mhaidat, Khaldoon, "Representations and Circuits for Time Based Computation" (2006). Scholar Archive. 1.