ARCNET Bus Models

ARCNET Bus Model

ARCNET Bus Delay Model

The delay model for an archnet bus computes the toke rotation time. This delay is the worst case delay for the bus. An introduction to ARCNET can be found here.

Figure_arcnettokenring : The ARCNET token ring protocol.

Figure_arcnettokenring shows the ARCNET token ring protocol. Node are logically organized in a ring with an order that depends on the address of the nodes. Before sending a message, a node needs to wait for an invitation to transmit (IIT) that is sent by the previous node in the ring. The, a node sends a free buffer enquiry (FBE) message and waits for an acknoledge form the destination node. The node, then. sends the packet whose length can vary and waits for the final acknoledge from the destination node.

The token rotation time can be computed knowing the nodes connected on the bus, the message period and the message length for each of them. This information is stored in the cosi::ThreadSet of each node. Each node on the bus can have multiple threads. Clearly, only sending threads (i.e. those threads with a non empty destination) contribute to the bus traffic. For a node, we can have the following cases:

A node is a source. In this case the node only has sending threads. To compute the worst case token rotation time, only the thread with longest messages needs to be taken into account.
A node is a destination. In this case the node only has receiving threads. This node contributes to the total delay but a token passing message only (i.e. the length of the IIT).
A node is a source of some messages and a destination of some others. This is the case for controllers that receive messages from sources (i.e. sensors), and send messages to destinations (i.e. actuators). To compute the worst case token rotation time, only the sending thread with longest messages needs to be taken into account.

Given a communication structure $ N(C,Q,L) $ capturing a bus, the bus delay model can compute the worst case token rotation time, and the worst case delay for a thread to access the bus. The model does not check that the communication structure is actually a bus. This must be ensured by the algorithm.

The token rotation time can be computed as follows:

$N \leftarrow 0$
for each arcnet node
1. let $\hat{l}(v)$ be the maximum message length among all threads
2. $N \leftarrow N + \textrm{ComputeMsgLength}(\hat{l}(v))$
return $N * T_{bit}$

The algorithm computes the total number of bits sent during a token rotation. This is worst case analysis because the longest messages only are considered. The total number of bits is divided by $T_{bit}$ which is the bit time compute as the inverse of the bust speed.

Procedure $\textrm{ComputeMsgLength}(l)$ computes the length of a bus message that is the sum of IIT, FBE, ACK, PAC and final ACK. The length of PAC depends on the parameter $l$ .

Generated on Sun Sep 7 18:37:45 2008 for COSI by

1.5.4

Contact

ARCNET Bus Models

ARCNET Node Model

ARCNET Node Cost Model

ARCNET Node Delay Model

ARCNET Bus Model

ARCNET Bus Cost Model

ARCNET Bus Delay Model