Maybe it can do with 50 ms latency with 3G/4G network? Thank you.
Impossible. You can't send a single packet with that little amount of latency over a mobile network, let alone capture video, encode video, mux video with audio, send it, receive it, buffer it, demux it, decode it, present it. 50ms latency per frame is not a whole lot higher than what you get with analog transmission!
You'll find that even many cameras on phones are going to have that much lag by the time the system gets the data to even work with it.
You realize it can take ~200ms for a human to even react to visual stimulus anyway? My TV takes at least 150ms to display a frame from its lossless HDMI input.
Your project requirements are completely out of touch with reality. You should also take time to gain an understanding of the tradeoffs that occur when you push digital video down into the extreme ends of low latency. You're about to make some real sacrifices by going under 1s or 500ms or so. Consider reading my post here: https://stackoverflow.com/a/37475943/362536 Particularly the "why not [magic technology here]" section.