z-wave mesh topography...

sorka

Active Member
So how does it work? With insteon, when a switch transmits a message, every switch within range retransmits the same message up to the repeat count (default is 3). Those switches, and the original retransmit the message on the next time slot(simplified explanation) so that they are all synchronized. This results in a stronger message that gets further on the next cycle allowing yet more insteon switches to recieve and repeat.

How does z-wave do it? Are they synchronized time slot repeaters in the way INSTEON does it over power line? Or is it single broadcast hoping. i.e. are there routing tables so that each z-wave devices knows that it's a repeater for other specific z-wave devices? And if it's that way, who sets up the routing tables?
 

Dean Roddey

Senior Member
The routing tables get set up when you add/remove modules from the network using the remote control. I assume that the modules themselves look for immediate neighbors. I don't know that for a fact. But the fact that you can send out a message to get a module to re-look up his neighbors would seem to indicate that this is the case. But they seem to be static otherwise, i.e. they don't seem to have any smarts to adjust for changes (hence the warning above about if somethign stops working.)

I would assume that Z-Wave's smarter brother (Zigbee) does handle such things in a more advanced way, but I again I don't know that for sure.
 

ChrisWalker

Active Member
The master controller sets up the routing tables, and devices can request updated routes. Additionally, central controllers can re-heal the network on the fly, much like an automatic defragmenter on your PC.

Chris
 
Dean Roddey said:
...But they seem to be static otherwise, i.e. they don't seem to have any smarts to adjust for changes (hence the warning above about if somethign stops working.)

I would assume that Z-Wave's smarter brother (Zigbee) does handle such things in a more advanced way, but I again I don't know that for sure.
It has been a while since I looked at the ZigBee 1.0 spec and AFAIK the ZigBee 2006/Pro spec is still not in general release outside the ZA... I'll send you some water softener salt if you need it.

The Z-Wave 1.0 spec specified a tree based network topology. I don't think there was a provision for branch pruning or merging. This resulted in "dissapointing" performance and most of the vendors of ZigBee products offered "ZigBee" or "What works" as part of their product offerings. This was the genesis of calling a product ZigBee if it contained the option of using the official ZigBee network topology in addition to "what works".

What I was told at the time, by a number of people, is that an official ZigBee network would loose any pretense at deterministic response at about 10 nodes.

What I understand today is that the 2006 spec includes a true mesh network topology and that the Pro spec includes some provision for end node devices to speak to each other directly.

In a conversation with Eric Smith at CES he indicated that there is some provision for Source Route Addressing for transitory and limited cases where a ZigBee 2006/Pro device finds itself talking to a ZigBee 1.0 device. I regret now that I didn't ask if this was part of the ZigBee spec or is a feature that Control4 implemented at the application layer for maintanance purposes.

So the overall answer seems to be: ZigBee 1.0 - Tree topology. Very little dynamic reconfiguration.

ZigBee 2006/Pro - True Mesh with dynamic reconfiguration (I have seen this demonstrated), Support for the older tree based topology and possibly some support for SRA.

HTH

George West
www.wtrs.net
 

Dean Roddey

Senior Member
What I was told at the time, by a number of people, is that an official ZigBee network would loose any pretense at deterministic response at about 10 nodes.

I don't know what the tolerance is for 'determinism', but I wouldn't expect a deterministic response necessarily from any wireless product where multiple modules are sharing the same transmission media without a token passing protocol of some sort.
 
Dean Roddey said:
What I was told at the time, by a number of people, is that an official ZigBee network would loose any pretense at deterministic response at about 10 nodes.

I don't know what the tolerance is for 'determinism', but I wouldn't expect a deterministic response necessarily from any wireless product where multiple modules are sharing the same transmission media without a token passing protocol of some sort.
Yabbut, "loose any pretense at deterministic..." sounds so much more classy than "falls over and dies." which was the actual quote.

George West
www.wtrs.net
 
Top