wireless - Multiple radio in a small dimensional sensor node PCB

I'm planning to develop a wireless sensor node that equips dual IEEE 802.15.4 radios. Although the board will equip homogeneous dual radios, both of them will operate at different channels. Regarding the design, however, I'm worrying about any possible bad correlations between the radios which will be placed in a relatively small dimensional sensor node such as 4" x 2".

1. 
   

   Is is possible to practically use dual radios in such a small dimensional board?

   
   


2. 
   

   If possible, what special care should I give the board in layout stage?

   
   


3. 
   
   

   If possible, I'd like to use a chip antenna in order to save space in a board. What is the pros and cons of a chip antenna? Can it be a proper decision? What special care should be taken for antenna selection?

   
   

Answer

If you want to use two radios, do they both need to be active at the same time? By this I mean could radio A be active for 30 seconds with radio B using the next 30 seconds for example?

Even better would be a single radio capable of operating on two frequencies. Most of the time it could be listening on both channels simultaneously and occasionally transmitting on one channel or the other.

This would greatly simplify your design as you would not need to worry about one radio interfering with the other.

If you really need for them both to be active (and perhaps transmitting) simultaneously this is possible, but would require good RF engineering and isolation of signal paths. On board chip antennas will greatly complicate the design, particularly if omnidirectional coverage is required. One thought would be to use a two sided board with each radio on it's own side and use a multi-layer board with one of the interior layers being a full copper ground plane for isolation.

EDIT: In response to "explain more in detail about why the chip antenna would bring about complicated hardware design "

If I were going to have two radios, one on each side of the board, I would think that a 1/4 or 5/8 wavelength antenna perpendicular to the surface(es) of the board would be a good choice with an inner layer of the multi-layer PC board serving as a ground plane for both antennas. Assuming the board was mounted in the horizontal plane then both antennas would have an omni-directional pattern with vertical polarization. Further assuming a sufficiently large ground plane then there should be maximum possible isolation between the radios.

You should be able to easily simulate such antennas using software like NEC2.