Search Results

This paper presents a novel encoding scheme as an alternative to Analog amplitude encoding for communicating sensor signals. The scheme has the potential of becoming a non-proprietary industrial standard for communicating sensor information to electronic control modules. Key features of the encoding scheme are the ability to communicate two sensor values using only 3 wires (power, ground and signal) with 12 bit resolution within 1ms. The scheme includes a checksum for error detection and a mechanism for reporting serial data such as low rate sensor information, part numbers or fault codes. Data is communicated to the receiving module by varying the time between discrete (single edge polarity) transitions. The encoding is self-calibrating and does not require an expensive crystal in the sending module (assumed to be a low-cost ASIC) to maintain signal integrity.

Currently in the automotive industry, most software source code is manually generated (i.e., hand written). This manually generated code is written to satisfy requirements that are generally specified or captured in an algorithm document. However, this process can be very error prone since errors can be introduced during the manual translation of the algorithm document to code. A better method would be to automatically generate code directly from the algorithm document. Therefore, the automotive industry is striving to model new and existing algorithms in an executable-modeling paradigm where code can be automatically generated. The advent of executable models together with automatic code generation should allow the translation of model to code to be error free, and this error-free status can be confirmed through testing. A three-stage process is presented to functionally verify the model, functionally verify the automatically-generated code, and structurally verify the code.

Recent advances in ANN (Artificial Neural Network) technology enable new methods to be developed in sensor technology. There are a large number of cases where there exists a causal relationship between one or more inputs and a physical quantity, but where an easily implemented analytical relationship between the inputs and the output can not easily be found. In such cases, machine learning techniques, such as artificial neural networks, are able to model that functional relationship. However, using conventional computing hardware, these methods, while theoretically attractive, are too computationally intensive for field deployment in real-time systems. Using a hardware implementation of an artificial neural network architecture, these computational restrictions can be eliminated.