So we'll look now at statistical multiplexing as well as resource pooling

as being two of the reasons why packet switching is very efficient.

Which in turn makes it scale-able.

So statistical multiplexing is the idea that multiple sessions can share one path.

Right, so if you consider just this again,

our simple one link with two sessions idea set.

even whenever anything on this pipe, right,

sender one and sender two can both use

the same link, right.

So, there's many different sessions that can be sharing the same path.

So, for instance sender two really doesn't need

a lot of this utilization right now, okay, so.

In time, as you see, as it's sending data, there's not very much.

And sender one can also send all of this, but

sender one can also fill in the blank spots here, right?

So suppose for instance that we're dividing this in time, right.

So this is one time slot, this is another time

slot, this is another time slot, this is another time slot,

right?

Under a C, under a a TDMA scheme, right,

where we're doing circuit switching this is sender one's time

slot, this is sender two's time slot, this is sender

one's time slot, and this is sender two's time slot.

But you see right here sender two only used half of its time

slot, right, and sender two here would have used none of its time slot.

Whereas that, that's very, that's like a wasted opportunity, so to speak.

So sender one can fill in those missing gaps, so to, quote, unquote, by,

you know, sending and then having the

network realize and then pooling that together.

So, we can think of this as a cocktail party analogy again, right?

So, if we're all in a room and we're standing in a room.

So if, if two pairs of people are speaking and they have their

time slot right now, but then they suddenly stop early and there's another

group of people that wants to talk,

right, what's to prevent them from talking, right?

So it's just more efficient because you're using all

of the resources, you're preventing a waste in resources.

The second one is resource pooling.

so in this case, one session can use multiple paths.

So statistical multiplexing means multiple sessions can use one path.

Resource pooling means one session can use multiple paths.

Right? So, here's this is one session.

Again, this is another session.

Right? So sender

two is sending. Right?

Again, sender two doesn't have a high demand.

Right?

So, if sender one was using this path and he had

already filled up the entire path with, with all of its data.

He could start sending some of his data down

to this path as well, or the network could.

I mean it wouldn't be sender one himself that

would make that decision, it would be the network.

But then he, then this link could start to take some of sender one's data.

Right? Under, again, under a circuit switch

scheme that, that can't happen because we have dedicated resources and

dedicated circuits, so we would be confined to one or the other.

But if this we're seeing here physically is links right, but really it's

the, it's the pour over on times frequencies and codes, and so on.

Just to say that anything that's wasted we can equalize the amount of demand

on each of the different resources, right,

by allowing people to share the resources instead.

So, in a cocktail party interesting analogy

we can make, right?

Now we're going to have more than one room, right, so,

to indicate the multiple different paths that we can use.

Right, so if one room is very crowded but we were allocated all to this one

room and everyone's talking, and maybe even there's

an equal number of people in each room, right?

Which means that the circuit switching scheme had allocated equal number

of resources to each person or equal, every person was basically

divided in the same way.

so then, but if in one room there's not much, there's

a lot of conversation going on and in this room everyone's silent

and everyone's shy for instance, what's to say that you couldn't

then come over into this room and have your conversation there instead?

Right?

So this would be the really loud and, room that had a lot of

congestion, and then you can move over and put someone in the soft room.

It's also similar if you think about a pool; just think about a literal pool.

And if you think about the way water fills up in a

pool, it's not just going to fill up in one specific location, right?

It's going to spread out to fill across the entire pool

so that every portion of the pool has the same level.

And that's really the demand level being

shared across all of the links, rather than

just pouring all the demand on to one link when the others aren't being utilized.

So, really what this does these two schemes, the bottom

line, there's, there's a lot of math that can go

into proving why statistical multiplexing has high efficiency, why

resource pooling has high efficiency, we won't do that.

But we're avoiding wasted network resources, that's the bottom line.

Statistical multiplexing and resource pooling

help us to avoid wasting resources.