Chapter 7: Wireless and Mobile Networks

Search this chapter

Audio Overview

0:00 / 0:00

Autoplay next chapter

Welcome to Last Minute Lecture!

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement not replaced the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Welcome everyone to the Deep Dive.

Just take a look around you right now.

Your phone, maybe your smart Lermostat, even the car you might have arrived in, they're all connected,

wirelessly.

It's really incredible how fundamental it's become, just woven into everything we do.

Totally, and think about the scale.

Back in, what, 1993,

there were like 34 million mobile cellular subscribers worldwide.

Yeah, a pretty niche thing back then.

Great, but by 2019, 8 .3 billion.

That's actually more subscriptions than there are people on earth.

That number is just staggering, and it shows how quickly this shifted from a luxury to, well, basically essential infrastructure.

Absolutely, but, you know, while we all use this stuff constantly figuring out how it all works,

how these devices connect and talk to each other in this complex networking world, that can feel pretty daunting.

It definitely can seem like a black box.

Oh, yeah.

So our mission today is to, well, simplify that complexity a bit.

This deep dive is really your shortcut to understanding the key ideas, the challenges, and the clever solutions in wireless and mobile networks.

And we're basing this on some solid ground.

We're digging into computer networking, a top -down approach.

The eighth edition by Kurose and Ross.

It's a fantastic resource.

Truly a foundational text.

Yeah.

And right off the bat, they make a distinction that's, I think, absolutely crucial for getting your head around this whole topic.

Which is?

The difference between wireless and mobility.

They aren't the same thing.

Okay, break that down.

Sure.

Wireless just describes the connection itself.

It's happening over radio waves, through the air, no physical cable involved.

Got it.

Like Wi -Fi or Bluetooth.

Exactly.

But mobility is different.

That's about the device's ability to move and change its point of connection to the network, maybe from one cell tower to another or one Wi -Fi hotspot to the next while staying connected.

Ah, okay.

So my smart fridge is wireless, but it's not really mobile.

My phone is both.

Precisely.

And understanding the difference wireless link versus mobile device is key.

So today, we'll unpack the theory, look at how it works in practice, maybe touch on some real world examples and connect it all to things like 5G, IoT, all that good stuff.

Sounds good.

So let's start with the basics.

The building blocks of any wireless network.

What are the pieces?

Okay, first you have your wireless hosts.

That's the end device, right?

Your phone, your laptop, but increasingly also all those internet of things devices, sensors, appliances,

even cars.

And like we said, they might be mobile or they might just sit there like a smart light bulb.

Exactly.

Then you have the wireless links.

That's the actual connection over the air.

Think phone to cell tower, laptop to your home Wi -Fi router.

And these links can be really different, right?

Like Bluetooth is short range, low speed, while 5G aims for huge range in speed.

Absolutely.

They vary a lot.

And managing those links often is the base station.

That's the central piece of infrastructure.

It could be a cell tower, could be a Wi -Fi access point, an AP.

Its job is basically to talk to the wireless devices and coordinate things.

Yeah.

Sending and receiving data, managing who gets to transmit when, that kind of thing, which leads to this idea of different mode.

Yeah.

Infrastructure mode versus ad hoc mode.

Most of the time you're in infrastructure mode.

Your device connects through a base station, like an AP, which connects you to the wider network, like the internet.

Okay.

That's like normal Wi -Fi at home or work.

Right.

But in ad hoc networks, there's no central base station.

Devices connect directly to each other.

I think maybe a couple of laptops sharing files directly using Wi -Fi direct or some battlefield communication systems.

They have to figure out routing and stuff themselves.

Interesting.

And when a device does move between base stations in infrastructure mode.

That's the handoff or handover.

Your phone switching from one cell tower to the next as you drive, for example, that's where the mobility part really kicks in and where a lot of clever engineering happens.

Right.

Keeping that connection seamless and all this wireless stuff eventually plugs into the bigger network infrastructure, usually the wired internet.

Correct.

And the book and our focus today is mostly on those common single hop infrastructure based networks, Wi -Fi and 4G, 5G cellular.

Okay.

So we have the pieces, but you mentioned challenges.

Why is wireless so much harder than just plugging in an ethernet cable?

It seems like it should be easier.

Yeah, you'd think,

but the airways are a much tougher environment than a nice shielded cable.

There are three big problems.

Okay.

Number one, decreasing signal strength or path loss.

Signals naturally get weaker as they travel, even in empty space.

And when they go through walls, furniture, people, they get attenuated sometimes significantly.

Yeah.

Everyone knows that feeling of the Wi -Fi dropping off at the far end of the house.

Exactly.

Problem two is interference.

Other things are also using radio waves, maybe another Wi -Fi network on the same channel, Bluetooth devices, microwave ovens.

They can all interfere with your signal.

I remember the old 2 .4 gigahertz cordless phones messing with early Wi -Fi.

That drove me crazy.

Oh yeah.

Classic example.

That's a big reason why newer Wi -Fi standards pushed into the five gigahertz band.

It was less crowded though.

Now it's getting busier too.

Okay.

Signal loss, interference.

What's the third one?

Multipath propagation.

This is a bit weirder.

Signals don't just travel directly.

They bounce off things, buildings, walls, the ground.

So the receiver gets the original signal plus slightly delayed copies or echoes.

Ah, so it sort of blurs the signal, makes it harder to read.

Precisely.

It smudges the signal, making it difficult for the receiver to decode the bits accurately.

So these three things, fading signals, noise, echoes, they all add up to what?

They lead to much higher bit error rates, BER, compared to wired links.

Just fundamentally, more bits get corrupted during transmission over the air.

Okay.

So errors are just a fact of life in wireless.

How do networks deal with that?

They have to build in robustness.

Unlike basic ethernet, wireless protocols like 802 .11 Wi -Fi use strong error detection, usually a CRC check, on every single frame.

And crucially, they use link layer acknowledgements and retransmissions.

So the receiver has to explicitly say, yep, got that frame okay.

Exactly.

If the sender doesn't get that ACK back quickly, it assumes the frame, where the ACK was lost or corrupted, and it sends the frame again.

It happens right there at the link layer, trying to fix errors before they even get up to higher layers like TCP.

Makes sense.

You need that reliability built in lower down.

And this relates directly to the signal to noise ratio, or SNR.

It's basically a measure of how your desired signal is compared to all the background noise and interference.

Higher SNR is better, right?

Easier to hear the signal.

Right.

Higher SNR generally means a lower bit error rate for a given transmission scheme.

But there's a trade off.

If you try to transmit data faster, using a more complex modulation scheme, you'll typically get more errors for the same SNR.

Ah, okay.

Speed versus reliability.

Exactly.

And this leads to a really clever feature in modern wireless, dynamic rate adaptation.

Your device and the base station are constantly monitoring the signal quality.

So they adjust the speed on the fly.

Yeah.

If the signal is strong and clear, high SNR, they'll try using a faster, more complex modulation scheme to push more bits per second.

But if the signal weakens or interference pops up, they'll drop back to a slower, more robust scheme to reduce errors.

So it's like your phone is constantly testing the waters, trying to go as fast as possible without dropping too many bits.

That's a great way to put it.

It's analogous to how TCP adjusts its sending rate based on congestion, but this is happening at the physical link layer based on the wireless channel conditions.

Super important for performance.

Okay.

Another big challenge you hear about is the hidden terminal problem.

What's that about?

It sounds sneaky.

It is a bit sneaky.

Yeah.

Imagine you have three devices, A, B, and C.

A, B is in the middle and can hear both A and C.

But A and C are too far apart or there's an obstacle so they can't hear each other.

Okay.

So A and C are hidden from each other.

Right.

Now suppose both A and C want to send data to B.

A listens, hears nothing because it can't hear C, and starts transmitting to B.

C listens, hears nothing, it can't hear A, and also starts transmitting to B.

Uh -oh.

Collision at B.

Exactly.

B receives two interfering signals at the same time.

I probably can't understand either.

And the problem is neither A nor C knew the other was transmitting because they couldn't hear each other.

This doesn't really happen in wired ethernet where everyone usually hears everyone else.

So that makes just listening before talking not quite enough for wireless.

Correct.

It complicates things significantly, which leads into how wireless networks actually manage access to the shared channel.

One approach, especially in older cellular and some other systems, is code division multiple access or CDMA.

Okay.

CDMA.

How does that try to solve the sharing problem?

It's a channel partitioning technique.

Instead of dividing the channel by time slots, TDMA, or frequency bands, FDMA, CDMA gives each user a unique code.

A code?

Like a secret password?

Sort of.

Think about it like this.

Each bit you want to send gets multiplied by this special faster changing code pattern, the chipping code.

Everyone transmits on the same frequency at the same time, but using their unique code.

How does that avoid interference?

The classic analogy is the cocktail party.

Imagine lots of people talking in different languages in the same room.

Even though you hear all the noise, you can tune into the conversation happening in the language you understand and filter out the others.

Ah, so the unique code acts like the language, letting the receiver tune in to the right sender.

Exactly.

The receiver uses the sender's known code to decode their specific signal, while other signals encoded with different codes just look like background noise.

It essentially partitions the code space.

Clever.

Are there downsides?

Yeah, you need carefully designed codes that don't interfere much with each other.

And ideally, the signals from different senders should arrive at the receiver with roughly equal power, which can be tricky to manage.

Okay, so CDMA is one way, but the technology most of us use daily for local wireless is Wi -Fi, right?

REC 802 .11.

Absolutely.

It's everywhere.

Homes, offices, coffee shops, airports.

Totally dominant for wireless LANs.

And it's gone through a bunch of versions, like 802 .11B, GN, AC, X.

Now they just call them Wi -Fi 4, 5, 6.

Right.

Each generation generally brought higher speeds, maybe better range, sometimes using different frequency bands, the older 2 .4 gigahertz, or the newer, often faster 5 gigahertz.

And the newer ones use fancy antenna tricks?

Yeah, things like MIMO, multiple input, multiple output, which uses multiple antennas on both sender and receiver to send multiple data streams at once,

or improve reliability.

And beamforming,

where the AP can sort of electronically steer the signal more directly towards your device.

Cool.

So, architecturally, how is a typical Wi -Fi network set up?

The fundamental unit is the basic service set, or BSS.

That's basically one access point, AP, and the wireless stations or devices associated with it.

And the AP connects to the rest of the network?

Usually, yes.

The AP plugs into a switch or router, connecting the wireless BSS to the larger wired network and off of the internet.

And just like wired devices, every wireless station in AP has its own unique 6 -byte MSE address.

So how does my laptop find and connect to the Coffee Shop Wi -Fi?

It uses a process called scanning.

The APs are constantly broadcasting beacon frames.

These are like little advertisements saying, hi, I'm here, my network name, SSID, is Coffee Shop Wi -Fi, and I'm operating on channel 6.

So my laptop just listens for these beacons?

It can.

That's called passive scanning.

Or it can do active scanning, where it sends out a probe request frame saying, anyone out there?

And APs in range will respond with probe responses.

Okay.

Then it sees a list of networks.

How does it choose?

Usually, it picks the one with the strongest signal, though that's not always best.

Maybe that AP is already overloaded with users.

Once it picks one, it goes through an association handshake, sending an association request, getting an association response from the AP.

And then it needs an IP address, right?

Yep.

Typically, it uses DHCP, just like on a wired network, to get an IP address, subnet mask, default gateway, DNS server info, all that.

Sometimes there's also an authentication step, maybe entering a password, like WPA23, or using more advanced methods.

Okay.

So once associated, how do all the devices connected to that one AP share the airwaves without constantly crashing into each other, especially with that hidden terminal problem we talked about?

Right.

This is where the Wi -Fi MC protocol comes in.

CS -Mac does.

Carrier Sense, multiple access with collision avoidance.

Okay.

CSMA Carrier Sense, it listens first.

But CA collision avoidance, not collision detection, like Ethernet.

Exactly.

That's the crucial difference.

Wi -Fi tries to avoid collisions in the first place.

Why?

Well, reliably detecting collisions in wireless is harder and more expensive, partly because signals fade, and partly because of the hidden terminal problem, you might not even hear the other station you're colliding with.

So if it can't easily detect collisions, it has to be more careful before transmitting.

Precisely.

And remember those high bit error rates.

Because of that, Wi -Fi requires link layer acknowledgments, ACKs.

When a station successfully receives a data frame, it waits a very short time, called SIF's short interframe space, and sends back an ACK frame.

And if the original sender doesn't get that ACK...

It assumes something went wrong,

the data frame was lost, or the ACK was lost, and it schedules a retransmission of the data frame, usually after waiting a bit longer.

Okay, so how does the collision avoidance part work?

What's the procedure?

It goes something like this.

One, if a station has data to send, it first listens to the channel.

Two, if the channel is idle for a specific period, called DS distributed interframe space, it can transmit immediately.

Okay, simple enough if it's clear.

What if it's busy?

Ah!

If the channel is busy, or becomes busy, during the DDS interval, the station doesn't transmit.

Instead, it chooses a random backoff value from a certain range.

A random waiting time.

Exactly.

It sets a timer with this random value.

The timer only counts down when the channel is sensed as idle.

If the channel becomes busy again while it's counting down, it freezes the timer.

And when the timer hits zero...

Then it transmits its entire frame,

and waits for the ACK.

And if it doesn't get the ACK...

It assumes a collision happened, or the frame ACK got corrupted,

increases the range from which it picks the random backoff value.

This is called exponential backoff.

And tries the whole process again.

So the randomness is key to stopping everyone from jumping in the moment the channel clears.

Yes.

That initial DTFS wait, plus the random backoff, even if the channel seems idle, is designed specifically to reduce the probability that two stations, both waiting for the channel, will start transmitting at the exact same time.

It's proactive collision avoidance.

Okay, that helps.

But what about the hidden terminal problem?

A and C still can't hear each other.

Good point.

For that, 802 .1 has an optional mechanism called RTS -CTS, which stands for Request to Send Clear to Send.

How does that work?

If a station, let's say A, wants to send a larger data frame to the APB, it can first send a very short request to send RTS frame to B.

Okay.

This frame basically says, Hey B, I want to send you some data.

Okay, what does B do?

If B is ready, it broadcasts a short, clear to send, CTS frame, back.

This CTS frame does two things.

It tells A, okay, you're clear to send.

And D, importantly, it tells everyone else within earshot of B, including the potentially hidden station, C, hey everyone, be quiet for this amount of time because A is about to transmit.

So even if C couldn't hear A's RTS, it can hear the AP's CTS and knows to shut up.

Exactly.

The CTS acts like a reservation signal broadcast by the central AP.

It helps silence potential hidden interferers for the duration of A's data transmission and the subsequent ACK.

So it makes collisions less likely, or at least makes them very short, just involving the RTS CTS frames.

Right.

It's particularly useful for large data frames where a collision will waste a lot of time.

In practice, whether it's used often depends on configuration.

There's usually a threshold.

And only frames larger than that threshold trigger the RTS CTS exchange.

Got it.

Now you mentioned the 802 .11 frame itself is interesting.

Something about addresses.

Yes.

This is quite unique compared to Ethernet.

An 802 .11 frame can have up to four MA address fields in its header, whereas Ethernet typically just has two, destination and source.

Four.

Why so many?

It's needed to handle the interaction between the wireless network, the BSS, and the wired network, the distribution system, DS, that the AP connects to.

Let's take the common case.

Your laptop, H1, sending data through an AP to a router, are one, on the wired network.

Address one is the MAC address of the immediate wireless receiver, in this case the AP.

Address two is the MAC address of the immediate wireless sender, your laptop, H1.

Okay.

Source and destination on the wireless link.

What's address three?

Address three is the MAC address of the ultimate destination on the wired side that this frame is heading towards, in this case, the router interface.

So the AP uses address three to know where to forward the packet onto the wired network after receiving it wirelessly.

Exactly.

It needs that final destination, MS, to create the proper Ethernet frame to send to the router.

When the router sends a packet back to your laptop, the addresses shift.

Address one is your laptop's MSC wireless destination.

Address two is the AP's MSC wireless sender.

And address three is the original source MMA address on the wired side, the router R1.

Address four is used in more complex scenarios like wireless distribution systems, where APs talk to each other wirelessly.

Wow.

Okay.

That's intricate, but makes sense for bridging wireless and wired.

It really is a clever piece of design.

The frame also has sequence number for tracking retransmissions, a duration field used by RTSCTS for channel reservation, and control fields for things like frame type and encryption status.

So what about mobility within the same Wi -Fi network?

Like walking down a hall past multiple APs connected to the same underlying wired network, same IP subnet, how does that handoff work smoothly?

Right.

The goal is seamless handoff, keep your IP address, keep your TCP sessions alive.

When your device decides to switch from AP1 to AP2, it associates with AP2.

But there's a problem.

The network switch that connects AP1 and AP2 still thinks your device's MSC address is reachable via AP1.

Its forwarding table is pointing the wrong way.

Exactly.

So here comes what the book calls a switch learning hack.

When your device associates with a new AP, AP2, AP2 sends an ethernet frame out onto the wired network with your device's MSC address as the source address.

Oh, so it pretends to be your device on the wired side.

Just for a moment with that specific frame, when the switch sees this frame coming from the port connected to AP2, but with your MSC address as the source, it updates its forwarding table.

It learns, ah, this AMEC address is now reachable via the port connected to AP2.

Future traffic for your device gets sent to the right AP.

That is a bit of a hack, but pretty clever.

It works.

And technologies like VLANs can extend this kind of same subnet mobility over larger physical areas.

Okay, switching gears slightly, but still related to Wi -Fi location finding.

We use acts like Uber, Waze, Maps all the time.

How do they know where we are, especially indoors where GPS might not work well?

Great question.

GPS is the baseline, of course.

Satellites broadcast signals, your phone listens to at least four, does some triangulation math, and figures out its latitude and longitude.

But as you said, GPS needs a clear view of the sky tough in dense cities or inside buildings.

So what's the alternative?

Wi -Fi.

This is the Wi -Fi positioning system, or WPS.

Companies like Google, Apple, Skyhook maintain huge databases mapping the AMEC addresses and SSIDs of millions, probably billions, of Wi -Fi APs worldwide to their estimated geographic locations.

How does my phone use that?

Your phone is constantly scanning for nearby Wi -Fi networks, even if you're not connected.

It sees the SSIDs and AMEC addresses of nearby APs and measures their signal strength.

It can then send this list of nearby APs and their signal strengths to a location service like Google's or Apple's.

And the service looks them up in the database.

Yep.

It looks up the known locations of those APs, considers the signal strengths, stronger signal probably means closer, and calculates an estimated position for your phone, often surprisingly accurately, even indoors.

But how did those AP locations get into the database in the first place?

Did someone drive around mapping them all?

That's part of it, especially initially with things like Google Street New Cars.

But here's the really cool crowdsourced part.

Your phone helps build and maintain the database.

When your phone does have a good GPS fix and it also sees nearby Wi -Fi APs, it can report back to the service, hey, I'm at this GPS coordinate and I see these Wi -Fi APs with these signal strengths.

Whoa, so by using the service, we're constantly refining the map of where all the Wi -Fi routers in the world are.

Exactly.

Thousands of phones passing by an AP over time, reporting its presence along with their GPS coordinates, help triangulate and refine that AP's estimated location in the database.

It's this massive ongoing collaborative mapping effort done almost invisibly.

That's genuinely fascinating.

A couple more advanced Wi -Fi things mentioned, weight adaptation and power management.

Right.

We touched on rate adaptation, dynamically changing the transmission speed based on signal quality.

The other big one is power management,

which is crucial for battery powered devices.

How does Wi -Fi save power?

An 802 .11 device can tell its AP, hey, I'm going to sleep for a bit.

The AP then buffers any incoming frames destined for that sleeping device.

The AP mentions in its regular beacon frames, which sleeping devices have buffered data waiting.

So the device just wakes up briefly to check the beacons.

Exactly.

It wakes up periodically, listens for a beacon.

If the beacon indicates data is waiting for it, it stays awake to retrieve it.

If not, it goes right back to sleep.

This can lead to huge energy savings.

The device might be asleep well over 99 % of the time.

Makes sense why our phone batteries last as long as they do, considering how much Wi -Fi they use.

It's a critical feature.

And that low power aspect is also central to another short range wireless tech, Bluetooth.

Ah, for headsets, keyboards, speakers.

Right.

Bluetooth is designed for personal area networks, pans or pecanettes.

Short range, low cost, low power, often acting as a cable replacement.

It operates in that busy 2 .4 gigahertz ISM band, the same one as older Wi -Fi and microwaves.

How does it avoid interference there?

It uses a technique called frequency hopping spread spectrum,

FHSS.

It rapidly hops between dozens of different channels in a pseudo random sequence known to both the sender and receiver.

If interference hits one channel, it only affects a tiny piece of the transmission before it hops away.

And it doesn't need an AP, right?

Correct.

It's typically ad hoc.

Devices form small networks, pecanettes, with one device acting as the master and others as clients.

Up to seven active clients per master.

They go through a discovery and pairing process to establish connections and synchronize their hopping patterns.

Okay, so that covers the local and personal area.

Let's zoom out to the wide area.

Cellular networks, 4G and now 5G, these give us that connectivity almost anywhere.

Indeed.

While Wi -Fi is great for homes, offices, campuses,

cellular provides that much broader ubiquitous coverage.

4G LTE really transformed things, making mobile HD video, payments, and tons of IoT applications practical.

And the key idea is cells, right?

Geographic areas served by a base station.

Exactly.

The network divides the landscape into cells, each with a base station, often called an enode B in 4G.

And a really important point about modern cellular 4G and 5G is that they have an all IP core network.

Meaning they use internet protocol for everything, even voice calls.

Yes.

It's a huge shift from the old circuit switch telephone network principles.

4G, 5G architecture embraces many core internet concepts, layering, separating data and control functions, using IP for transport end to end.

So let's look inside that 4G LTE architecture.

What are the main components?

Okay.

Broadly, you have the radio access network at the edge and the core network called the EPC or Evolved Packet Core.

Edge and core, familiar concepts.

Right.

Key elements.

Your mobile device or user equipment, UE.

Then the base station, Enode B, managing the radio link.

In the core network, you have several important pieces.

Like what?

On the data plane, handling your actual traffic are the serving gateway, SGW, and the packet data network gateway, PGW.

The PGW is the gateway to the external internet.

It often assigns your phone, its IP address, maybe via NAT, and is the anchor point for your connection.

Okay.

So data flows, UE, Enode B, SGW, PGW, internet.

Roughly, yes.

But then there's the control plane, the brains managing the operation.

The key component here is the Mobility Management Entity or MME.

It handles things like authentication, setting up the data paths and tracking your device's location.

So the MME isn't actually forwarding your data packets.

Correct.

It sets things up but stays out of the high speed data path.

It works with another control element, the home subscriber server, HSS, which is a database in your home network holding your subscription info, authentication keys, etc.

You mentioned the MME sets up data paths.

How does that work?

It uses tunnels.

When you connect, the MME orchestrates setting up IP tunnels, specifically using GTP, two -price tunneling protocol, between the Enode B and the SGW, and between the SGW and the PGW.

Your actual IP packets are encapsulated inside these GTP tunnels as they traverse the core network.

Why tunnel?

What's the benefit?

Mobility.

When you move from one Enode B to another, the MME just needs to tell the SGW to update the endpoint of the tunnel from the old Enode B to the new one.

The SGW and PGW don't necessarily need to change, making handovers smoother.

The tunnel hides the mobility deeper within the network.

That makes sense.

And you said the MME tracks location.

Yes.

At least down to the group of cells you're currently in.

This is needed for things like paging if your phone is idle, sleeping, and a call or data comes in.

The MME needs to know which group of base stations to tell, hey, wake up this device.

It's amazing how we got here.

The book mentions the evolution from 2G, 3G.

Yeah, it's a fascinating journey.

2G was circuit switch voice.

3G was a hybrid.

It added a parallel packet switch data network alongside the existing voice network and the bolting data capabilities on.

4G was the revolutionary step to a unified all -IP core network, where everything, including voice, as voice over LTE or VoLTE, is treated as packet data.

It wasn't an overnight switch, but a phased evolution.

And below the IP layer on that radio link between the phone and the base station, there are specialized LTE protocols.

Yes, several sub -layers handle things specific to the wireless link.

The Packet Data Convergence Protocol, PDCP, does IP header compression and encryption.

The Radio Link Control, RLC,

handles segmenting large IP packets and provides reliable data transfer using acknowledgments.

And the Medium Access Control, MAC layer, handles scheduling which devices get to transmit when.

So more layers to handle the specifics of wireless before you even get to IP.

Right.

And remember those TTP tunnels in the core?

They encapsulate the user's IP datagrams for forwarding between the gateways identified by Tunnel Endpoint IDs, TAIDs.

How does LTE actually manage the radio spectrum itself?

How does it allocate bandwidth?

It uses a technique called OFDM, Orthogonal Frequency Division Multiplexing.

You can think of it as dividing the available radio channel into many narrow sub -channels, frequency division,

and also dividing time into slots, time division.

So it's like a grid of time and frequency.

Exactly.

And the base station, ENO -B, acts as the scheduler.

It dynamically allocates specific time frequency slots within this grid to different mobile devices based on their needs, their channel conditions, and network policies.

It often uses opportunistic scheduling, trying to give slots to users who currently have good signal quality to maximize overall throughput.

And LTE Advanced allowed combining channels.

Yeah.

Carrier aggregation lets devices use multiple frequency channels simultaneously for even higher bandwidth.

Okay.

Two more LTE functions.

Connecting to the network and saving power.

Right.

Network attachment is a multi -step process.

First, your phone scans for signals and attaches to a base station.

Then there's mutual authentication involving the MME and the

to verify both the device and the network.

Finally, the MME orchestrates setting up those data plane tunnels we talked about.

ENO -B to SGW, SGW to PGW.

And power saving, similar to Wi -Fi.

Similar ideas, yes.

LTE has sleep modes.

Discontinuous Reception, DRX, is a light sleep where the device and ENO -B agree on short periods when the device will wake up to listen for pages.

Idle state is a deeper sleep with much less frequent wakeups.

If the device moves cells while in deep sleep, the MME uses that paging mechanism to find it when data arrives.

Looking globally, all these cellular networks connect, right?

Absolutely.

Your home carrier connects to other carriers for roaming and to the wider internet, often through specialized IP exchange, IPX networks.

It really is a network of networks, much like the internet itself is a network of ISPs.

Which brings us to the latest generation, 5G.

What are the big promises there?

Oh, the ambitions are huge.

We're talking ubiquitous multi -gigabit speeds, extremely low latency, down to one millisecond target for some applications, and the ability to support a massive number of simultaneously connected devices.

What kind of applications need that?

Think high quality AR -VR,

truly autonomous vehicles that need instant communication, remote surgery,

real -time industrial automation, maybe even replacing fixed broadband in some areas, fixed wireless access.

What are the key improvements over 4G?

The targets are roughly 10x higher peak bitrates, 10x lower latency, and supporting maybe 100x the traffic capacity or device density.

How does it achieve that?

Does it use different radio waves?

Yes, that's a big part of it.

5G uses existing cellular bands, called FR1, but also opens up much higher frequency bands known as millimeter wave, millimeter wave, or FR2.

These MME frequencies offer huge amounts of bandwidth, enabling those gigabit speeds.

But there is a catch with millimeter waves, right?

There is.

They don't travel as far as lower frequencies.

And they're more easily blocked by obstacles like walls, buildings, even rain or foliage.

So deploying 5G, especially the inland wave version, requires many more smaller base stations placed close together compared to 4G.

And 5G isn't just one size fits all.

Correct.

It's designed around three main use cases, often described as coexisting standards.

One, EME, Enhanced Mobile Broadband.

This is about faster data speeds for phones, tablets, hotspots, think better video, AR -VR.

Two, UR -LLC, Ultra Reliable Low Latency Communications.

Focused on applications needing near instantaneous response and very high reliability like factory robots, vehicle -to -vehicle communication, remote control systems.

Three, MMTC, Massive Machine Type Communications.

Designed to support huge numbers of low power, low data rate devices like sensors, smart meters, asset trackers, the IoT explosion.

Needs to be very power efficient.

So how does 5G actually boost capacity so much?

You mentioned more base stations.

Right.

Increased cell density is one factor.

More available spectrum, especially with the inland wave, is another.

And the third is higher spectral efficiency, getting more bits per second out of the same amount of frequency.

This comes from using even more advanced MIMO antenna techniques, like massive MIMO, with potentially hundreds of antennas at the base station, allowing it to create very narrow beams and talk to many users simultaneously in the same frequency band.

Multi -user MIMO, or EMU MIMO.

And the core network changes too.

Significantly.

The 5G core network is designed from the ground up to be cloud -native and service -based.

Functions are virtualized, running as software, allowing for more flexibility and deployment closer to the edge.

For example, the user plane function, UPF, handles packet forwarding and can be distributed near the network to reduce latency.

Control plane functions like Session Management, SMF, and Access Mobility Management, AMF, are also separated and virtualized.

Okay, that's a lot on the wireless links and network structures.

Let's put it back to Mobility Management.

How does the network handle devices physically moving around while staying connected?

Right.

This is the core network layer challenge of mobility.

We're not just talking about turning your phone on in a new city.

We're talking about moving during an active connection, say, on a phone call or streaming video, and having the network seamlessly reroute your data to your new location.

That process is the handover or handoff.

And this relies on the idea of home versus visited networks.

Yes.

Every mobile device has a home network, your cellular provider.

When you travel outside your provider's direct coverage, you connect to a visited network, another carrier with a roaming agreement.

Your home network acts as the anchor point, the permanent record holder for your subscription via the HSS.

So the visited network needs to talk to your home network to let you connect.

Exactly.

They need to coordinate for authentication and billing.

Think of your home network like your permanent mailing address.

Even if you travel, people can contact your home base to find out where you are or get mail forwarded.

Which brings us to how data actually gets to a mobile device, the correspondence dilemma.

How does someone on the internet sending you data know your current IP address if you're moving around different networks?

Great question.

One theoretical way would be for every mobile device to constantly advertise its specific current IP address into the like BGP.

But that's completely unscalable.

Billions of devices constantly changing routes.

The internet's routing tables would explode.

Yeah, that sounds like a nightmare.

So what's the practical solution?

Push the mobility handling to the edge, specifically using the mobile device's home network as an intermediary.

The main approach is called indirect routing.

How does that work?

The correspondent, the device sending you data, doesn't need to know where you are.

It just sends the data packet to your permanent IP address, which is associated with your home network.

Okay.

So the packet arrives at your home network's gateway, like the PGW and LTE.

What happens then?

The home gateway intercepts the packet.

It then queries the HSS or equivalent mobility database to find out which visited network you're currently connected to and what your temporary care of address or tunnel endpoint is in that visited network.

So the home network knows where you're roaming.

Right.

Then the home gateway takes the original packet,

encapsulates it, puts it inside another IP packet address to the visited network's gateway, and tunnels it across the internet to the visited network.

Like sending a letter inside another envelope.

Exactly.

The visited gateway receives the outer packet, decapsulates it to get the original packet, and then forwards the original packet to your mobile device using its current local address on the visited network.

So the correspondent sending the data is completely unaware of all this tunneling and Correct.

It's transparent to the correspondent.

The return path from the mobile device back to the correspondent can either tunnel back through the home network or sometimes use local breakout to go directly from the visited network to the internet, if allowed.

This indirect routing is the basis for mobility in both mobile IP and 4G, 5G.

What are the downsides?

The main one is potential inefficiency known as triangle routing.

If you're roaming in, say, London, but your home network is in New York and you're talking to someone else also in London, the data might go London, New York, London, and back again.

That seems wasteful.

It can be.

An alternative is direct routing.

Here, the correspondent first queries your home network, HSS, to find your current location, visited network.

Then the correspondent itself tunnels the data directly to your visited network, bypassing the home network for the data packets.

That sounds more efficient.

Why isn't it always used?

It adds complexity.

The correspondent needs to implement this location query mechanism.

And what happens if you move while the correspondent is sending data?

The correspondent needs a way to be notified of your new location and update its tunnel.

Indirect routing is simpler in many ways, even if potentially less optimal.

OK, so focusing on the practical implementation in 4G, 5G, which uses indirect routing, how does a handover actually happen step by step?

Say I'm on that video call and walk from cell A to cell B.

Right, let's simplify it.

Your phone is connected to base station A, the source.

As you move, the signal weakens, or maybe base station B to target, offers a stronger signal.

The source base station A decides a handover might be needed, often based on signal strength measurements reported by your phone.

Source A sends a handover request message to target B, usually via the MME.

This request includes info about your connection.

Target B checks if it has resources, radio slots, etc.

to accept your connection.

If yes, it preallocates them and sends a handover request acknowledge back to source A.