Future work: I am interested in new projects and can operate directly or via a professional services entity. 

Previous work: Here is a set of work artefacts (papers+patents list, some presentations, etc.). The remainder of this page gives an overview of R&D projects.

Current work is in highlighted in section Control and Signal Processing.

Control and Signal Processing

• • Next Generation GPS ground stations and receivers: I am working with Locata Corp. Canberra on the next generation of their advanced mm accuracy GPS system as used in ports and mines for autonomous vehicles. We are currently increasing the computing power available by using some of the latest reconfigurable hardware and programmable RF systems. Here is the previous generation Locata GPS systems in operation on robotic cranes at the Port of Auckland, where they had been in use in support of human driven cranes. Current use is in mines and ports in USA and Europe e.g. Port of LA has over 140 autonomous robotic cranes in operation in early 2025. (2018-)

  • Guide Star Laser: I worked with Dr James Webb on modern control applications for laser stabilisation as part of the Guide Star Laser project at SERC (2016-2018). We performed real-time modelling and controller design to maintain the phase locked states of two light beams in a closed path with mixing in a non-linear prism, and with one mirror under real-time control. The output power of interest is at 589nm which coincides with optical excitation of sodium in the upper atmosphere, and is the basis of forming a target on which ground based telescopes can focus. Lab tests achieved continuous operation for up to 45 minutes and powers of the order of 8W output. This work is part of the photon pressure strand of SERC Program 4 (R1.4, search for Musumeci).

  • Industrial Printer: I have led small industrial R&D teams developing Ultra High Speed Continuous Paper Inkjet Printers in two startup companies, and was co-founder of Champion Imaging Systems (Melbourne AU).

    • Phillip Musumeci, MultiLayer Continuous Paper Printer, International Patent WO0112444, publication date 22-Feb-2001 (various jurisdictions).

    • A seminar "Multi layer printer project" describes the design challenges, team roles, and includes an example of complex state machines doing autonomous data delivery (just a set of finite states machines interacting with other, triggered by the outside world, and with no dependency on executable code when "running").

    • Other info:

      • a 6-layer printer used a "Form Flo" modular collator and achieved 200 feet/minute printing in Melbourne;

      • a 1-layer printer was presented at the Dusseldorf 2000 DRUPA trade show and used a Dimuken paper transport system (high speed paper rewinder); 

      • alternative copy of WO 0112444 on Google Patents (with links to Chinese and other versions, and citations).  

      • This work developed the fastest inkjet printers at the time (an order of magnitude speed increase was obtained).  The core team members were: Evan Greensmith#, David Burn, Simon Farrugia, Phillip Musumeci, and at the pointy end when things had to work: James Webb#.   (# later completed a PhD)

      • Tools: C++ (BSD Unix), real-time C (AVR uP, TI C6000 DSP), Xilinx (schematic entry), VHDL, circuit layouts (design and review).  

  • Modern Control Applications: I worked with Bob Bitmead in the ANU Systems Engineering Department on design and implementation of optimal control algorithms to control speed of steam turbines driving sugar milling units. This work led to better quality of control from less actuator use, allowing the plant to run more productively. This collaborative work involved ANU Systems Engineering and Bob Bitmead (now at UCSD Mech and Aerospace Eng), the CSR sugar mill at Ingham (also the site of trials), and was funded by CSR.

    • Phillip Musumeci and Bob Bitmead, LQG Control of Sugar Cane Crushing, ANU and CSR Report, 1990.

    • This work introduced new algorithms to control in sugar mills, and in a broader context gave confidence to others willing to attempt modern real-time control. Other details in this interview with Bob Bitmead at the time Systems Engineering merged with Computer Science to form RSISE.

    • Tools: Matlab (control systems and system ID toolboxes), real-time f77 (oh yes!), Significant algorithms == {Kalman filtering, LQG control, recursive least squares}.

  • Agricultural Machinery: My MEngSc (Automatic control of base butter height on sugarcane harvesters, JCU Electrical Engineering, supervisor Bob Bitmead) investigated height sensing algorithms and sensors for sugarcane harvesters. Techniques investigated included: front mounted sonar; rear mounted post-cut surface roughness sensing; and variance measurement on the hydraulic oil pressure drop across the motor driving the base cutters.

    • P.C. Musumeci and R.R. Bitmead, Basecutter Height Sensing on Cane Harvesters, Proceedings of Australian Society of Sugar Cane Technologists, 1981.

    • Tools: Sonar systems used an HP 21MX minicomputer managing transmission and reception of orthogonal sequence modulated acoustic energy, hydraulic oil pressure analysis used a harvester mounted custom Motorola microprocessor based data logger.

Automated Design and Language Translation

• • Notes on Systems Design Work for IT systems that can automatically design and build new IT systems (including, ideally, better versions of themselves). (2007)

    • Tools: C (minimalist OS service access on BSD&linux), python (exploring behaviours to optimally allocate registers), custom graph language (designing higher level language features), mediawiki (team documentation and communication).

  • Semi-automated Program Translation: A Fortran to C++ translator has been developed in python to translate a 17500+ LOC train timetabling application. It features an iterative approach to learning about data types, and preserves all "common" block data structures as new classes. Most code structures of the original program are retained so that existing staff knowledge remains useful. Improved code quality by introducing a class to provide physical model parameters and a class to centralise physical model and general run-time exception handling (replacing hard-coded constants scattered in the source code, and use of subprogram multiple returns). Testing used scripts to directly compare and verify all generated timetable and driver instruction data for existing mining and metropolitan lines. It is interesting to discover that modern Fortran compilers can and do use data flow analysis to determine data types, and they can also support multiple language dialects (versions) in the same code body although this last feature does not assist translation. Client: Constraint Technologies Melbourne (search www.constrainttec.com for TPAC Rail Timetable). (2011/2012)

    • Tools: python with libraries such as {re regular expressions, StringIO, pickle, pprint, md5 tagging}, C++ with Boost libraries, small C code wrappers for classic stdio C sources (Apple BSD). Learning algorithms were implemented in a state machine with memory of source code patterns.

Ad-hoc Networks

• • Ad-hoc Networking/Pervasive Computing: Colin Lemmon (JCU IT discipline) and I have worked on packet routing in Wireless Ad Hoc Networks:

    • Boundary Mapping and Boundary State Routing (BSR) in Ad-Hoc Networks, Colin Lemmon and Phillip Musumeci, IEEE Transactions on Mobile Computing, vol. 7 no. 1 January 2008 (abstract);

    • Cooperative Behaviour of Location Aware Nodes in Ad-hoc Networks, Colin Lemmon and Phillip Musumeci, IEEE-DEST 2007, February 2007, Cairns, Australia. (abstract) (refereed)  

    • We have also investigated geographic coordinate data updates (with data compression) and trust propagation in Ad Hoc Networks. I found that route discovery and path planning ideas from mobile robots work was very relevant to this field.

    • News February 2010: Colin has completed his PhD Boundary Mapping and Its Application to Geographic Routing (JCU).

    • Tools: C++ (custom tools).

Creative Industries

• • • Built Environment: Ben Guy (Urban Circus), Nic Bidwell (JCU IT discipline), and I have considered how games engine technology might apply to city planning:

      • GamePlan: Serious Gaming for Place Making, Ben Guy, Nicola Bidwell, and Phillip Musumeci, 2nd Australasian Conference on Interactive Entertainment, November 2005, Sydney, Australia. (ACM portal) (refereed)

    • Regional Creative Industries: As a member of a creative industries working party within the cluster group program of the Cairns Regional Economic Development Corporation, survey questions were developed to measure current (2006) CI activity. Analysis by Cummings Economics led to the report The Ideas Economy: Creative Industries - their size, worth and impact in Far North Queensland in May 2008.

      • In common with overseas studies, this work found many unexpected sectors to be design based.

Image and Array Processing

• • Robotics: Genevieve Coath (RMIT) and I have worked on image analysis algorithms for RoboCup. Some of our work to produce a RoboCup vision system that was not dependent on known object colouring is described in:

    • Adaptive Arc Fittingfor Ball Detection in RoboCup, Genevieve Coath and Phillip Musumeci, Australian Pattern Recognition Society, WDIC2003 (Brisbane February 2003). ISBN: 0-9580255-2-5 (refereed)

    • News May 2007: Genevieve has completed a PhD in the Dynamic Systems and Control Group, Mechanical and Manufacturing Engineering, Uni of Melbourne.

    • Tools: octave, C.

    • Movie: RMIT 2000 vs. Italian Golem team (honours student team project that preceded this image analysis work, shows actual robots).

  • Optimal Filtering and Array Processing: My PhD (Optimal Array Processing for Seismic Inversion, ANU Systems Engineering, supervisor Darrell Williamson) developed 2-d/3-d array processing algorithms and 1-d seismic inversion algorithms for use in tomographic analysis of earth structures. Selected papers:

    • Fast Design of Optimal Array Filters, Phillip Musumeci, IEEE International Conference on Acoustics, Speech, and Signal Processing, Adelaide, Australia, pp.V509-V512, April 1994. (PDF) ISBN 0-7803-1775-0;

    • P. Musumeci and D. Williamson, Optimum Two-Dimensional Array Filters, Proceedings of the 1989 Australian Symposium on Signal Processing and its Applications, Adelaide, Australia, pp.373-376, April 1989;

    • D. Williamson, P. Agathoklis, and P. Musumeci, Discrete Modelling for Solving One-Dimensional Inversion on Finite Media, IEEE CDC Proceedings, Austin, Texas, USA, pp.1892-1896, December 1988;

    • D. Williamson, K.L. Teo, and P.C. Musumeci, Optimum FIR Array Filters, IEEE Trans. Acoustics, Speech, Signal Processing, vol. ASSP-36, no. 8, pp.1211-1222, August 1988;

    • This work developed optimal array filter structures that could be efficiently designed on-the-fly while suppressing noise over a minimum number of spatially distinct signal samples in order to obtain a "clean" signal from a narrow spatial region of interest (i.e. a probed path), and then an analytic solution to the lossy wave equation was discretised as the basis of an efficient deconvolution algorithm to discover properties on this probed path (i.e. wave propagation "resistance" and the related velocity which then changes the path estimate...).

    • Tools: Maple, Matlab, NAG and NSWC maths libraries.

  • Image Coding: I have worked in a team performing research on coding algorithms that efficiently handle digital image sequences at different (but related) spatial resolutions from HDTV to videophones. This was useful because it allowed better conversion between formats and researchers became more familiar with the techniques employed e.g. motion compensation, subband coding, Huffman and run-length coding, etc. This collaborative work involved ADFA, Telstra, Siemens, with Australian Government support, and was based at ADFA, Canberra.

    • J. Arnold, M. Cavenor, P. Musumeci and C. Sholl, "Coding Scheme for Flexible Interworking of Video Services on B-ISDN", 4th International Workshop on Packet Video, Kyoto, Japan, August 1991.

    • This pre-MPEG2 work tested candidate algorithms as a contribution to the MPEG committee goals.

    • Tools: Matlab, C, custom extensions to Matlab (C).

Microprocessor/Real-time Systems

• • Application areas include mechanical harvesters (AusToft, QLD), underwater environmental monitoring (prototypes, AIMS, QLD), electrical energy use logging (govt. generator, QLD), automotive body flexing analysis (Ford, VIC), audio compression and embedded unix (Trio Datacom, VIC), fire station monitoring (CFA, VIC), reptile monitoring (JCU, QLD), farm soil and environment monitoring with zigbee networks (Land & Water Management, QLD).

  • Tools: various uP/uC devices (8 and 16 bit), integer DSP devices, real-time assembly language and C.