|
On an EIB network, the bulk of run-time communication passes between sensors, controllers and actuators on a peer-to-peer basis using multicast group addressing. This results in one-to-many messages, which are extremely bandwidth-efficient and ensure typical reaction times in the 100 ms range (on the EIB Twisted Pair medium).
publisher-subscriber model
A distributed application with EIB multicast binding may loosely be regarded as a set of shared variables. A device may publish some particular datapoint as a shared variable on some given group address. Any other device on the EIB system may hold a local copy of this variable (referred to as a group communication object), subscribing to the same group address; this copy will be kept in sync whenever the publisher updates its value.
It is clear that multiple publishers may use the same group address, to achieve n-to-m coupling.
grouping at dapatoint level
As pointed out, the "grouping" of target entities which may be reached with a single message is done at the level of individual datapoints; this approach is much more flexible and powerful than grouping targets merely at the node or device level.
flexible binding
By means of this principle, very flexible and powerful binding (i.e. linking published and subscribing datapoints by assigning group addresses) is possible at the level of individual variables or datapoints.
Using binding tools such as EIBA's PC-based ETS, actual binding may optionally be defined at project design time only; this is sometimes referred to as late binding. In this way, the person designing the project achieves substantial command over the customization and behaviour of the application in each single case.
| 
