DSP & Algorithms

Algorithms come in many forms, and for many purposes; from digital signal processing of data from some form of sensor, highly mathematic computations for calculation, to cryptography including encryption, authentication and data integrity checking.

A typical algorithm development path for a digital signal processing application might be:

  • Definition of the use cases for the system/function under consideration.
  • Modelling of system inputs (typically in Matlab) and/or collection of sample input data.
  • Creation of initial algorithmic models (typically in Matlab).
  • Initial algorithm implementation in floating point C on prototype target for demo.
  • Further data collection and algorithm improvement.
  • Target implementation of algorithm in floating point C, with MEX wrapper to enable direct comparison/validation against original Matlab model.
  • Algorithm optimisation if required (dependent on application and target):
    • Reducing computation complexity (typically including fixed-point C implementation including performance versus complexity trade-offs; often required for low-power operation
    • optmisation for processing speed (for time-critical operation)
    • optimisation to minimise ROM/RAM use (typically for implementation on a low-resource target)

Examples of Edenprime’s algorithm design and implementation work include:

  • Analysis of seismic signals for detection of nearby activity with source-type discrimination (algorithm development and fixed point implementation)
  • Implementation of optimised cryptographic authentication functions on ultra-low resource target to meet time and resource constraints
  • Analysis of motion and position sensor data for detection of significant fault events in critical infrastructure (algorithm development and fixed point implementation)
  • Implementation of Digital Pre-Distortion for high-speed antenna waveforms.
  • Optimised fixed-point logarithm, exponential and geometric function libraries on low resource target
  • Optimised exponential curve matching algorithm on low resource target

Scroll to Top