The requirement is to build a sensor network using several solar/battery powered nodes, of which, only one has connectivity with the outside world (GPRS). Requirements:
Node-to-node LOS distances 1Km and up.
Acceptable frequencies: 868MHz or 2.4GHz.
Topology: point-to-multipoint or preferably mesh.
Form factor & interface: Xbee-socket-compliant & serial-interface
The devices wake-up simultaneously and periodically (i.e. every 15 minutes), send data to the GPRS-capable device and fall back to sleep.
An obvious solution is to use Digimesh Xbees, since in Digimesh all devices can fall to sleep. The problem with that is that Digimesh runs on series-1 Xbees, which are not that good in terms of power-efficiency and range. On the other hand, if using Zigbee Xbees, the coordinator should not fall to sleep. BTW, does anyone has a clue of the time it takes for a Xbee-Zigbee network (i.e. one coordinator and 10-15 end-nodes) to be ready for sending and receiving data from the moment the coordinator and end-devices are turned on?
We would like to avoid using simpler RF modules (i.e. ciseco XRF/ARF) and had to develop MAC/networking SW for these.
There are some really good RF-modules that could do the job (i.e. ZWIR4512, ST SP1ML) but would like Xbee-socket format and proven solution.
The devices wake-up simultaneously and periodically (i.e. every 15 minutes), send data to the GPRS-capable device and fall back to sleep.
Simultaneously is going to present a problem. How will one device know that another is waking up? Why flood the collector with all the data "simultaneously"?
An obvious solution is to use Digimesh Xbees, since in Digimesh all devices can fall to sleep.
The obvious solution, to me, is to use Series 2 (mesh network) models. Any XBee can go to sleep. That is not a function of how they are connected.
Line of sight distances of 1 km or more could be challenging. Models that can reach that range are power hungry beasts.
Well, the problem is that the collector is battery operated as well and thus cannot be awake all the time. The devices have real-time clocks, thus it is easy to arrange that all devices wake-up the same time.
We could set it up so that the collector wakes up twice or three-times more often than end-nodes and collect half or one-third of data each time but i believe that battery-wise it will be worse for the collector, which is the most critical node anyways (battery wise).
I have not any significant experience with Xbees but I thought that Series-2 modules support only the Zigbee protocol and that in a Zigbee network, the coordinator and router-nodes are not supposed to go to sleep. Do you believe that if we put the coordinator into sleep as well we won't have any problems (i.e. long join times)?
The devices have real-time clocks, thus it is easy to arrange that all devices wake-up the same time.
Um hmm. As long as plus or minus a couple of minutes meets your definition of "at the same time".
I thought that Series-2 modules support only the Zigbee protocol and that in a Zigbee network, the coordinator and router-nodes are not supposed to go to sleep.
I have no idea why you thought that. Of course, I have no experience with putting XBees to sleep or waking them up, so I have no idea how long they need to be awake before the network is established.