Selection of an appropriate wireless network topology

Dear Sirs,

I'm considering now my final year project on electronics eng. The project is a wireless indoor network of inexpensive low-power devices, that have to communicate with a main unit, connected to PC. Following is the description of this network:

1) The number of network devices is big, say 1000-2000 2) There is no a need that the devices talk one to each other. The only necesarry communication link is between such a device and the main unit. So the topology that looks suitable for my application is the Star topology, but I have a lot of doubts regarding the possibility to implement this topology for 2000 devices. 3) I would like that the both kinds of devices will have a possibility to initiate a communication each to other, but if this feature is complicating the protocol, I'm ready to renounce that only Master will initiate the communication. 4) Broadcasting and multicasting possibility: the PC must have a possibility to turn to all devices or to a group of devices 5) Radio Coverage problem. I thought about repeaters, but if I have a lot of doubts abou this, for example,if there installed a few repeaters and say 3 repeaters received the same message, which one of them will retransmit the message ? and etc. Another solution is building of another kind of network, where instead of the star topology, every device will talk to the closest to him device, so I will not need repeater, but the problem is how to build a path from one device to the main unit ?

Since I see the issue is very complicated, I thought to look for any off-the-shelf solution: either to use a designed software stack or moreover, to use a chip, if exists. I have considered the SimpliciTI protocol, but it is intended for a small RF networks (less than 100 nodes). Now I'm learning the Wirelss Sensor Network, but I fill it is complicated too. So, my question to the forum is anybody aware of any existing solution for my problem ? Do you think the task is too hard for the final year project, even if I will find a partner, and it is better to give up with it now ? A few years ago, on my work, I have succesfully designed and implemented (firmware) combimed wired-wireless network, so I have a bit understanding and experience with wireless networking, but this network seems to me as from a higher league.

Thank you in advance for every suggestion/recommendation.

Reply to
elil1975
Loading thread data ...

Yes.

Reply to
linnix

Are you required to design a new one from the ground up? If no -> google "mesh network module" "dust mesh network" If yes -> pick a new project

Reply to
larwe

You need to specify the physical environment (distance between master and slaves, what kind of obstacles etc.) the data rate required and the update rate.

If you can use 4-8 RF-channels, you could use any protocol with only a single byte slave address and use the slave address numbers on each frequency. This will also increase the polling rate, since only 250 slaves needs to be polled at each frequency. The Rx/Tx turn turn around time can reduce the throughput in half-duplex networks.

If you are using multiple frequencies on the master site, the transmitter might block a receiver on the adjacent site, if the transmitters and receivers are located on the same site, so plan putting the transmitters and receivers at separate sites or at least into different vertical polarized antennas at different hights in the mast.

Alternatively, use a different site for each frequency, so the same antenna is used for both Rx and Tx at each site. Using a specific frequency for each slave and you could use a directional antenna at the master site. Use a radio link at a different frequency or a cable (RF, RS422 or ethernet) to carry the signal to the main PC.

Paul

Reply to
Paul Keinanen

...

...

He need to specify the budget. Assuming that he can build each for $10 to $20. Using $20K to $40K for a student project is unreal. The money is better use for hiring another professional student, if the OP is rich.

"doubts on feasiblity" confirmed "permission to withdraw" granted

Reply to
linnix

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.