REQ: Linear topology wireless network.

[Ichan]

Anyone?

I am thinking to control / monitor many solar powered lighting pole wirelessly, HOW?

Not with GSM, any info appreciated, TIA.

-ichan

[Bobbla]

I am thinking to control / monitor many solar powered lighting pole wirelessly, HOW?
any info appreciated, TIA.

Well, thats not a lot of information, BUT I would suggest that maybe you'd want to look in to zigbee or some 802.15 based device?
Also, I know that you can get MCU's with integrated RF from both ATMEL and Microchip.
Are there any cost, IO, power, range, bandwidth, security or other requirements?

Some links:
http://en.wikipedia.org/wiki/Wireless_personal_area_network
http://en.wikipedia.org/wiki/Comparison_of_802.15.4_radio_modules Real nice tables
http://en.wikipedia.org/wiki/Comparison_of_wireless_data_standards Nice tables here as well..
http://en.wikipedia.org/wiki/Zigbee

edit: just noticed I wrote charts, changed to tables... sorry about that.

[hate]

What will be the range by 'wireless'? Is it like one node is in another country than the other or are they kilometers apart or the solar is in your garden and you want to control it from your house etc...? For the latter case I can recommend Microchip's MRF49 for a custom protocol (you have to implement it) transceiver or XBee with a standard protocol. If you provide more details, I might have better suggestions.

[Ichan]

I should add "Street" before "Lighting Pole" on my question, the distance between pole is about 20 - 50 meter, the number of node (pole) can be more than 1000 but 255 node per network coordinator is acceptable. The data rate requirement is very low, several bytes data + command per node per day is enough. Security is not important, cost is always  .

I read some about Zigbee and I think I have some modules lying around for quite some time but had no occasion yet to play with. The link to the module comparison is very interesting, thanks, but still i am not sure which one to choose.

Microchip MiWi Pro seems good and the stack is free but i can not find any reference related with linear implementation, can a mesh network devices implemented in linear configuration?

Waiting for suggestions.

-ichan

[hate]

So will your maximum distance be 50 meters or is this the maximum distance between 2 nodes? From what I understand former is not the case as over 1000 lighting poles in say 50 square meters is unlikely. If I'm right, you have to go for a high power transceiver, I've seen XBees in various power ranges and in the basic mode they act as simple serial connection which you can easily adapt to your project by creating a command frame consisting of node id, command, checksum. Otherwise if you need a maximum range of 50 meters, you may opt to go for mrf49 solution as it is not that hard to work with and costs much more less than the XBee.

[solutions]

Disagree on the high power transmitter requirement. I believe Zigbee nodes can pass on info not intended for them. With  a light pole, a latency of several seconds over 1000 nodes is no big deal in the application since it is one way from the sounds of it.

[hate]

I've tested an XBee with a 2mW transmitter rating. It could barely transfer the signal through one single wall. I don't expect a transmitter of this power to broadcast over 1km range if need be. That was what I meant for high power transmitter in case I couldn't express myself clearly.

[Bobbla]

the distance between pole is about 20 - 50 meter, the number of node (pole) can be more than 1000 but 255 node per network coordinator is acceptable. The data rate requirement is very low, several bytes data + command per node per day is enough. Security is not important, cost is always  .

1. The way I understand it is that you have a series of light poles you want to control/monitor.
N1 --- N2 --- N3 --- N4 --- N5 --//-- Nn   ...(Nx - Node number x)
What happens if node 4 dies? will you be able to communicate with N5? or higher?
If your wireless system is able to communicate with more nodes then the two closes to them, there can be room for failure. (Redundancy good, failure bad)

2. Will you be able to have antenna's? (Omni-)Directional?
Can you influence the design of the poles in such a way that cheap (Adjustable?) directional (patch?) antenna's can be attached? Are there any obstacles in the way?

3. According to the zigbee.org's FAQ it as a 64k network size, not 255. http://www.zigbee.org/Specifications/ZigBee/FAQ.aspx#12

4. Are there special requirements for interfacing with the network with humans or machines?

5. I forgot..


edit:

I believe Zigbee nodes can pass on info not intended for them. With  a light pole, a latency of several seconds over 1000 nodes is no big deal in the application since it is one way from the sounds of it.

As I understand it, the zigbee will pass stuff on as part of a mesh network. Seems like the application requires two way communication to me, monitor and commands...

[Ichan]

1. The way I understand it is that you have a series of light poles you want to control/monitor.
N1 --- N2 --- N3 --- N4 --- N5 --//-- Nn   ...(Nx - Node number x)
What happens if node 4 dies? will you be able to communicate with N5? or higher?
If your wireless system is able to communicate with more nodes then the two closes to them, there can be room for failure. (Redundancy good, failure bad)

2. Will you be able to have antenna's? (Omni-)Directional?
Can you influence the design of the poles in such a way that cheap (Adjustable?) directional (patch?) antenna's can be attached? Are there any obstacles in the way?

3. According to the zigbee.org's FAQ it as a 64k network size, not 255. http://www.zigbee.org/Specifications/ZigBee/FAQ.aspx#12

4. Are there special requirements for interfacing with the network with humans or machines?

5. I forgot..

1. Very good point, i will take a note for this - redundancy is needed.

2. Yes, it should have external antenna, small whip antenna like on XBee Pro is what i think about, omni directional it is. The device will be positioned on the top of the pole so line of sight can be taken as default, but perhaps on some location there will a big tree or ad billboard or large road sign that can block the transmision - some kind of repeater / beacon probably needed on this situation. Another note to take.

3. I have to read more about Zigbee, but i guess on linear configuration it can not be that much.

4. No requirement for that.

5. No, the power source will be taken from the pole PV system (battery) - Yes, if the pole power died then the node too, more reason for point #1.

-ichan

[solutions]

I've tested an XBee with a 2mW transmitter rating. It could barely transfer the signal through one single wall. I don't expect a transmitter of this power to broadcast over 1km range if need be. That was what I meant for high power transmitter in case I couldn't express myself clearly.

It sounds like you have star on the brain. First rule of problem brainstorming is to let go of your prejudices. You also missed a key element in the problem definition;

"distance between pole is about 20 - 50 meter"

Assume one failed node, that says 100m, line of sight. There are no walls between streelights where I live - can't be sure about that one where you are.

The foliage could be an issue, but is nothing a backup satellite link wouldn't solve.

[hate]

Well I wasn't talking about any topology. What I had in mind was the simple master and slave scenario. The reason for that is the number of nodes in question. Supposing they are over 1000 and I think linking them in sequence is not the best way, in other words not a good way at all. And for a reasonable network of this size, the kind of protocol should be TCP-IP, even v6 might do good. ;p Speaking of probability theory, the probability of failure for 2 consequent nodes is [(1/1000)*(2/999)] excluding the end nodes. That makes about 1/(5*10^5) and in communication, data channels with error rates higher than 1/(10^6) are NOT even considered secure and by secure I don't mean someone else can break in or not. So yes I think the transmitters should be high powered to cover more than many nodes to repeat the signal to many other nodes if need be.

Btw, the kind of walls that go between street lights here are the houses themselves. We usually have parallel and perpendicular streets surrounding the houses and you can pick 2 street lights each on a different parallel or perpendicular street. I can provide a map for you if that's not clear enough.

[Parmin]

NRF24L01+ devices may fit the bill.
I have not explore the device fully yet but I believe daisy chaining command can be arranged.

[Bobbla]

2. Yes, it should have external antenna, small whip antenna like on XBee Pro is what i think about, omni directional it is. The device will be positioned on the top of the pole so line of sight can be taken as default, but perhaps on some location there will a big tree or ad billboard or large road sign that can block the transmision - some kind of repeater / beacon probably needed on this situation. Another note to take.

3. I have to read more about Zigbee, but i guess on linear configuration it can not be that much.

So mesh networks do not need to know about everyone else in the network. http://www.freaklabs.org/index.php/Blog/Zigbee/zigbee-mesh-routing-interactive-tutorial.html

Assume one failed node, that says 100m, line of sight. There are no walls between streelights where I live - can't be sure about that one where you are.

Should note that you can get zigbee devices that will have different power output/ranges.. XBee has from 1mW to 1W.. depending on what type you choose.

Btw, the kind of walls that go between street lights here are the houses themselves. We usually have parallel and perpendicular streets surrounding the houses and you can pick 2 street lights each on a different parallel or perpendicular street. I can provide a map for you if that's not clear enough.

Most streetlights have wires/cables hanging between them, so line of sight between them is usually not a problem. But no scenario is given here really... so an open mind is required. That said, a mesh network can ignore a lot of nodes if it is more efficient to bypass them all together. Example, instead of going up and down a street to send a message to the next street, if one node in the next street is visible, then we simply ignore the first street and send directly to the 2nd street.

I would also suggest you might look into using both short range and long range zigbee devices:
Say.. short range will let you talk to 2-3 (100m) nodes forwards, and long range will let you talk to maybe 10 (500m) nodes forward. Then you can sorta mix in area's where you need to, if you experience any trouble, this will give greater coverage and redundancy options.

Nordic Semi... aren't they in the short range business??

[hate]

Most streetlights have wires/cables hanging between them, so line of sight between them is usually not a problem. But no scenario is given here really... so an open mind is required. That said, a mesh network can ignore a lot of nodes if it is more efficient to bypass them all together. Example, instead of going up and down a street to send a message to the next street, if one node in the next street is visible, then we simply ignore the first street and send directly to the 2nd street.

I would also suggest you might look into using both short range and long range zigbee devices:
Say.. short range will let you talk to 2-3 (100m) nodes forwards, and long range will let you talk to maybe 10 (500m) nodes forward. Then you can sorta mix in area's where you need to, if you experience any trouble, this will give greater coverage and redundancy options.

If you are suggesting cable transmission between the nodes that don't have direct line of sight between them, I agree it will be a more robust solution if available. Most of the cables are buried underground here though.

Whether short range or long range, the nodes shouldn't be linked in sequence imo. They should cover other nodes than just the next couple of nodes.

[Gallymimu]

The Contiki mesh network stack is pretty nice (http://www.contiki-os.org/) we are using it with a cc2538 SoC for some ultra low power mesh monitoring systems.  If you have a large per unit budget, Linear has some AWESOME mesh modules using a protocol called Wireless HART.  They are easy to use but pretty pricey (they want $30 per CHIP in qty 1000).

Zigbee is a pretty good choice if you have power available (like in OPs scenario).  Zigbee is useless though for a dispersed battery based network since it is a star topology rather than a true uniform mesh.

[Ichan]

The contiki seems too complex for this app, but the cc2538 is indeed one of the choices - TI has Z-Stack for cc25xx family which is free to download after some registration.

IEEE 802.15.4 (including Zigbee) seems the right one, and for the sake of cost i think it should go to transceiver level rather than module. Up to now the choice is narrowed to TI CC25xx with Z-Stack and Microchip MFR24J40 with MiWi Pro.

Zigbee is useless though for a dispersed battery based network since it is a star topology rather than a true uniform mesh.

Would you please elaborate the words in blue above? Is that in comparison with "Digi Mesh"?

Mesh Networking seems applicable, but still i can not find anything about linear implementation. Linear topology that i asked, because the pole installation is lined up, aiming street lighting pole for highway or toll road - not residential street lighting.

-ichan

[Ichan]

I decided to go experimenting with MRF24J40, the MiWi Pro stack is easier to digest. 

One more question come: is 2.4 GHz the right choice for this?

-ichan

[mike_au]

Lower frequency will carry further (and through more obstacles) for a given power but won't be as fast (the exact details of the obstacles, terrain, etc. can have an impact but this is a pretty safe general rule). Since you are talking relatively long distance and low bandwidth, something like 900, 433 or 27MHz might be a better option (assuming it is part of the unlicensed band with similar transmit power limits to 2.4GHz where you are). The downside is that 2.4GHz is much more popular (at least when talking about hobbyist electronics) so it is easier to find high-ish power radios, antennas, etc. in 2.4GHz than it is in other frequencies.