Papers

PDF papers

Technical Reports

 

Counter Representation in Microprocessors

In this paper we consider some counter designs with a very small amount of logic. We start by considering common designs - base two, Gray codes and linear feedback shift registers. For an n-bit counter, we take a look at counters of cycle size 2^n −1 and find that we can often save one XOR gate over a linear feedback shift register. We then investigate designing an n-bit counter for a particular cycle size, and discover that cycles of size 1 . . . 2^n −1 are much easier to find than cycles of size 2^n . We finish by looking for the smallest possible 2^n cycle counters.

The UNDERDOG, current design and development

The UNDERDOG (Under Deep Ocean Glider) is an autonomous underwater gliding robot. It has a target glide depth of 3000m and a crush depth of 4500m all in a package that weighs a little under 30kg. The hull is comprised of three modular units that are interconnected for the glider format, however they can also be utilised separately for individual deep sea applications. The structural integrity of the hull at depth is validated using the COSMOSWORKS finite element analysis software. We have also built a test tank capable of testing real components at pressures up to 300 Bar to test non structural potential failure points such as seals and feedthroughs.

An inertial wave-powered generation system for small ocean science platforms

This paper describes our initial investigation into an inertially coupled wavepower generation system, designed for use in small, low-power, long-endurance ocean exploration robots. The system recovers small amounts of power from movements of the robot caused by ocean waves. The primary focus of the investigation to date has been the development of efficient power recovery system electronics that extract useful power from a range of ocean wave amplitudes and frequencies, and thus a very wide range of generated voltages and currents.

NEC2++: High-Performance Numerical Electromagnetics Code

We describe NEC2++, an open-source numerical electromagnetics code compatible with NEC-2. This object oriented C++ code, derived from the original FORTRAN, provides an extremely high performance, readable, modular foundation for research into numerical electromagnetism and antenna simulation. In addition, the NEC2++ code has a C API and can be compiled into a library for inclusion into other software – for example automatic antenna design optimisation systems, plugins for other languages, and graphical front-ends. We present an example of the use of NEC2++ for genetic antenna optimization.

N-Body Gravitational Simulation Using Dedicated Hardware

There any many computational problems in which the time complexity for a simple algorithm is O(N 2 ), but with more complex algorithms, can be reduced to O(N.ln(N )). Examples of these are: the n-body problem, ray- plane intersections for ray-tracing, interactions between neutral molecules modelled using the Lennard-Jones potential, and sorting objects in a list. Some of these faster algorithms, like the tree algorithms used for the n- body problem, become not only more complex but, can also reduce the accuracy of the result, even if only slightly in many cases. An alternative to using more complicated algorithms is to make the simple algorithms very, very fast. The solution presented uses a Field Programmable Logic Array (FPGA) to calculate the force due to gravity between two bodies at 100 million times per second. This is with only a single, unoptimised calculation pipeline, with future versions expected to run at 200 Mhz, and two pipelines per one million gate FPGA.

A hardware framework for application specific supercomputing

Talk at the IEEE Workshop on High Performance Computing, Auckland, 2006

A CPLD Coprocessor for Embedded Cryptography

Microcontrollers and CPLDs (Complex Programmable Logic Devices) are both low power components commonly found in embedded devices. We consider how a microcontroller and CPLD can work together to perform encryption and decryption with better power vs. performance characteristics than either device could manage alone.

Efficient Hardware Calculation of Inverses in GF (2^8)

Galois Fields are an important in many areas including cryptography and error correcting codes. We consider various approaches to calculate multiplicative inverses of elements in the Galois Field GF (2^8) in hardware, as is used in the Rijndael encryption algorithm. We compare speed vs. implementation area for these approaches.

A Temperature Controlled CMOS Camera

In recent years CMOS imaging sensors have been replacing CCD imagers in consumer and industrial markets. Their low cost and high peripheral integration on chip gives new opportunities to those who have not adopted CCD technology due to cost or technical concerns. This paper describes the design of a prototype CMOS camera for use in spectral experiments with noise reduction techniques to improve image quality. We use a Peltier device to cool the sensor, an SDRAM frame buffer for storage and a USB interface to transfer the data at high speed.