Radio networked devices

[markp]

Hi,

I am currently looking at a design that involves a very low cost radio based
'pseudo' network of devices that can be up to 2km apart.

Basically the requirement involves having up to 64 battery powered devices,
and a central controller. The controller must be able to address and send
data to each device individually or via broadcast, and the devices must also
be able to send data back to the controller when they need to. There is no
need for the devices themselves to inter-communicate.

The solution I'm currently looking at involves a time division multiplexed
radio link using one single channel. Timeslots are used to prevent
transmitters on each device from interfering with each other at the
controller's receiver, as they could all possibly want to communicate at the
same time, and the controller also has a timeslot for transmission so it
does not get interfered with either.

The requirement is complicated somewhat by the devices being battery
powered, and therefore cannot keep their receivers powered up continuosly
due to excessive power consumption. It's therefore important to be able to
power-down and power-up both transmitter and receiver sections of the
transceiver quickly. I also have the problem of synchronisation which I
believe can be done using this system.

Price is also real issue here, the transceiver must be as cheap as possible
(preferably < £10 each), and use a harmonised band such as 433MHz or 868MHz.

Can anyone suggest a very cheap solution to this?

Thanks!

Mark.

 

[John Larkin]

Hi,

I am currently looking at a design that involves a very low cost radio based
'pseudo' network of devices that can be up to 2km apart.

Basically the requirement involves having up to 64 battery powered devices,
and a central controller. The controller must be able to address and send
data to each device individually or via broadcast, and the devices must also
be able to send data back to the controller when they need to. There is no
need for the devices themselves to inter-communicate.

The solution I'm currently looking at involves a time division multiplexed
radio link using one single channel. Timeslots are used to prevent
transmitters on each device from interfering with each other at the
controller's receiver, as they could all possibly want to communicate at the
same time, and the controller also has a timeslot for transmission so it
does not get interfered with either.

Sounds tricky. Why not have each field device have a unique address
and operate in slave-only mode, transmitting only in response to a
command message from the master?

Variants include having each unit power down for some commandable time
interval after it's polled, and allowing break-in slave transmissions
if something unexpected happens at a remote site. Lots of collision
resolution algorithms are floating around.

John

 

[hamilton]

Hi,

I am currently looking at a design that involves a very low cost radio based
'pseudo' network of devices that can be up to 2km apart.

<snip>

Hitting a remote device thats 2Km from the transmitter is hard enough
without adding muliple units. ( and bi-directional to boot !! )

Try looking at "Wireless Micro-networking" or "ac-hoc wireless embedded
network".

I had this link for a while and think this may be a better solution.

http://www.jlhlabs.com/LowPowerWireless.htm

Once you start looking at this, it gets better.

good luck.

 

[markp]

Sounds tricky. Why not have each field device have a unique address
and operate in slave-only mode, transmitting only in response to a
command message from the master?

Variants include having each unit power down for some commandable time
interval after it's polled, and allowing break-in slave transmissions
if something unexpected happens at a remote site. Lots of collision
resolution algorithms are floating around.

John

Hi John,

Explicit polling is an interesting option, and would certainly be easier to
implement. The TDM approach is very similar though, but the polling slots
are self timed and the polling is essentially automatic without the need for
the controller to transmit. The bandwidth is not too high from these units
so this could be a viable option. One of the reasons for wanting to do this
is to allow firmware upgrades of the remote units from the controller, and
to be informed when a device stops functioning or is tampered with, but they
all do need to communicate reasonably quickly with the controller when they
detect a trigger condition, so sending a poll command between each poll will
just waste time. There's no reason why both modes can't be done though, TDM
for fast polling and explicit polling/addressing for other functions.

Thanks!

Mark.