Sharing Real-Time Objects in Distributed Embedded systems

Student thesis: Master programme thesis

  • Erland Ketil Larsen
  • Ib Havn
4. term, Master of IT (Continuing education) (Continuing Education Programme (Master))
With increasing intelligence in consumer products, distributed embedded systems have become
more common. Sharing real time objects between nodes are one of the major challenges in cost
effective systems as they often lack the resources needed for technologies like RMI and
CORBA. This report describes a real industrial case of a resource constrained distributed embedded
system, and suggests a middleware that could improve performance and maintainability
of the product.
The contribution of this project is to implement an effective middleware and communication
layer that will ease the development of distributed applications in small resource constrained
embedded systems. CORBA Event Service and “The ACE ORB” (TAO) is studied, and optimisations
are suggested to decrease complexity and code size. For communication services, the
Local Interconnect Network (LIN) is studied as inspiration for implementation of a transparent
communication layer using the standard UART embedded in most microcontrollers, and RS-
485 hardware on the physical layer.
The designed and implemented middleware is divided into separate modules with very little
coupling, this makes it possible to replace the communication service to match other systems
communication technologies and bus structures. The modularity of the middleware makes it
event possible to use the middleware on a standalone node without the communication service
An API is provided, and a simple example of a distributed application is programmed on four
8-bit Atmel microcontrollers connected by a RS-485 bus.
The demo system shows that it is relatively simple to implement a distributed embedded system
with the provided API, and the latency of the communication service is predictable.
A low coupling to hardware makes it easy to port to another platform and/or another network
SpecialisationSoftware Construction
Publication date27 Mar 2015
Number of pages99
ID: 209323386