For wired communication we have protocols like as CAN,Modbus,... between devices,What is a small protocol for wireless communication for 8 bit micros like as AVR.
Good question. 'Zigbee' is a standard protocol, that allows multiple devices to communicate with eachother in a network, with a medium datarate. Its similar (but not exactly) to Bluetooth in its principle.
I don't know any 'standard' protocol for microcontrollers. Are you planning on building a device that can communicate with many different devices? Or just a point-to-point link? Because, creating something that can cope with many transceivers in a network can be very tricky. But if you just wish two micro's to 'talk' to each other wirelessly then things become simpler. Look out for 'keeloq' in microchips appnote section. Also, search this forum for 'manchester encoding' a, in my opinion, its a good way to get data over radio relatively safely. As for a 'packet' protocol, its pretty simple and depends on your app. Address byte, databyte etc...
To be honest, I've been struggling trying to come up with a protocol for my wireless system. Mainly because many systems are really compllicated and I don't know why, makes my system look 'amateur'. If you just have one transmitter and one receiver (or two transceivers) then as long as both know the protocol, things should be good.
I've used these, and they're just like using wired rs232 (well, UARTs anyway). They're designed to do almost everything for you and replace a 'cable'. There are many similar products on the web if these are not available to you. Plus, there are lower-end modules that are just data radios (a lot cheaper), which means you'll have to encode any data, DC-balance it, and probably add error detection - not overly difficult, but a hassle none-the-less.
These sort of, 'data modems' are becoming increasing popular, purely because they'e so easy to use, and they make for a reliable data link.
Mesh networks are a great concept. But the challenge lies in managing the dynamics of mesh networks so users receive an acceptable level of performance in terms of both latency and throughput.
4- The clients at the last rung will receive -at best- less than a *millionth* of the total bandwidth at the root. The reason: each of the 25 clients at each hop are sharing the bandwidth available at that level equally, in a true ad hoc mesh network. And with each hop the radio that provides access to the next level has its bandwidth reduced hy half. So the poor guys that the 4th rung get 1/25 * 1/2 * 1/25 * 1/2 .. you get the message.