Other (Change?) | Consumer Products

Can GigE Vision deliver on its promise?

GigE Vision is, in theory, a highly attractive option as a bus standard for implementing machine vision systems. It offers high specifications for data rate and interconnection length, promises interoperability between vendors, and is based on a familiar, mainstream technology with all the concomitant advantages of economies of scale and ubiquity of deployment.

 

However, GigE Vision is also a relatively young standard that is yet to be proven in the broad range of machine vision applications.

Sony’s Image Sensing Solutions division, in conjunction with the Institute of Photonic Microsystems (IPMS) of the Fraunhofer Institute, undertook a study, seeking to objectively compare the specifications and performance of GigE Vision, with that of the most-common current machine vision interconnection standard, IIDC (Instrumentation and Industrial Digital Camera) over IEEE 1394.

This whitepaper outlines the methodology and key results of that research.

The study concludes that although GigE Vision (when supported by the GenICam specification) does indeed provide better performance than IIDC/IEEE 1394b in terms of raw data rate and maximum cable length, the more recent standard nevertheless contains gaps and shortcomings that make it less suitable for use in machine vision implementations. In particular, because GigE Vision provides no fundamental quality of service (QoS) guarantees, it lends itself less readily to the transmission of real-time data. Performance can also be compromised by the fact that GigE Vision tends to produce more CPU loading than IEEE 1394, encumbering the host processors within the system.

Moreover, the combination of IEEE 1394 and IIDC allows integrators to work via an API (application programming interface) at a high level, whereas GigE Vision/GenICam requires the designer to be actively involved in low-level configuration tasks, for instance setting packet size: as a result GigE hardware is more difficult to integrate, more difficult to optimize, and exhibits restricted interoperability between vendors.

The study concludes that GigE Vision exhibits great theoretical promise for machine vision applications, but is still evolving. Real-life systems therefore may not, in practice, deliver all of the benefits and performance enhancements that designers expect, and many designers will choose – for the moment at least – to stick with established architectures rather than risking a move to new technology.

Send your request to zone@eu.sony.com to get a copy of the full white paper.

Bookmark with