I will keep it short, say you have 4 perspectives, each with the same setup in 1080p60 @ 2500Kbps, combining in to a 4K video at the ingest server.
Now, your ingest server needs to map each perspective on to a 4K canvas and upload that stream somewhere, let's assume a 5 Mbit 2160p60 stream this is what you get:
Clients 2.5 x 4 = 10 Mbps upload.
IServer 10 Mbps download.
IServer 5 Mbps upload.
EDIT: Twitch caps @ 6 Mbps for streaming so it doesn't matter if 5 Mbps is enough or not, 6 is max for Twitch.
Now, your ingest server needs to map each perspective on to a 4K canvas and upload that stream somewhere, let's assume a 5 Mbit 2160p60 stream this is what you get:
Clients 2.5 x 4 = 10 Mbps upload.
IServer 10 Mbps download.
IServer 5 Mbps upload.
EDIT: Twitch caps @ 6 Mbps for streaming so it doesn't matter if 5 Mbps is enough or not, 6 is max for Twitch.