Image Processing Reference
In-Depth Information
For the TP medium, a unique medium access mechanism labeled
predictive p-persistent CSMA
is used. Its key mechanism is that when confirmed multicast services are used, a certain prediction
on the future network load (i.e., the confirmations to be expected) can be made. The length of the
arbitration phase is modified accordingly. hus, the rise of the collision ratio with increasing load is
mitigated. This helps to insure an acceptable minimum packet rate even under heavy load, unlike
in Ethernet-style networks using CSMA/collision detection (CD), where the network load has to be
kept well below %. At the start of the arbitration phase priority time slots are available for urgent
messages.
The entire routable address space of a CNP network is referred to as the
domain.
Domains are
identiiedbyanIDwhoselengthcanbechosenuptobitcorrespondingtorequirements(asshort
as possible, since it is included in every frame; as long as necessary to avoid logical interference,
especially on open media). A domain can hold up to
subnets
with a maximum of nodes each.
Hence, up to , nodes can be addressed within a single domain. A subnet will usually correspond
to a physical channel, although it is both possible for multiple physical channels to be linked into a
subnet by bridges or repeaters as well as for multiple subnets to coexist on the same physical seg-
ment. Routing is performed between different subnets only. In particular, domain boundaries can be
crossed by proxy nodes only (which transfer the information on the application layer). Subnets are
usuallyarrangedinatreehierarchytoallowtheuseofself-learningrouters.
Every domain can host up to multicast groups. Groups can include nodes from any subnet.
Broadcasts can be directed to a single subnet or the entire domain. Each node carries a worldwide
unique -bit identification, the
Node ID
. It can be used for addressing individual nodes for manage-
ment and configuration purposes, while regular unicast communication is handled through logical
subnet and node addresses.
For both unicast and multicast, a reliable transmission mode (
acknowledged
)withend-to-
end acknowledgments can be selected. In addition to the “one-shot”
unacknowledged
mode, an
unacknowledged-repeated
mode is provided, where every transmission is automatically repeated a
fixed number of times. When acknowledged multicast is used, groups are limited to a maximum of
members each. Otherwise, they can contain an arbitrary number of nodes from the domain.
For acknowledged transmissions, a challenge-response authentication mechanism is provided.
The challenge consists of enciphering a -bit random number using a -bit shared secret.
The CNP application layer allows generic application-specific messaging, but offers particular
support for the propagation of network variables. Network variables are bound via -bit unique
identifiers (
selectors
). The management and diagnostic services include querying the content type
of the network variables (self-identification), the node status, querying and updating addressing
information and network variable bindings, reading and writing memory, device identification, and
configuring routers.
Thanks to the open specifications, second source protocol implementations exist besides the
Neuron chip. Moreover, variants with integrated transceivers are available. A derivative of ANSI C
called Neuron C is used to program the Neuron chips, whereas standard ANSI C can be used to pro-
gram other controllers. Figure . shows the classic architecture of a LonWorks node. he controller
executes the seven OSI protocol layers and the application program, which interfaces with sensors
and actuators connected through the I/O interface.
A variety of installation and management tools are available for CNP networks. he wide majority,
however, is based on the LNS (LonWorks Network Operating System) management middleware by
Echelon. Besides APIs for commissioning, testing, and maintaining, LNS provides a common project
database, avoiding vendor lock-in of these valuable data. For configuration of vendor-specific param-
eters LNS provides a plug-in interface. Using LNS is not compulsory, but ensures open access to the
project data (which represent a considerable value in engineer's work).
Applications exchange data via network variables. These are defined within the nodes' applica-
tion software. During system configuration, logical links are established between network variables
