# Using a modem in a LAN to access the Internet

Summary:
Understand the use of a modem to connect to an ISP and to the internet
Hello,

I have been thinking about this and would like to se if anyone can provide some validation or corrections:

The internet is a huge network interconnecting smaller networks. The smallest network would be classified as a LAN (local area network). WANs are instead big networks of networks. Different networks that are part of the Internet are interconnected using routers and high data transmission channels. But a router can also connect computers inside the same network....

In the case of a small home network, which is a LAN, the various computers/machines have their own local IP address an communicate with each other using the Ethernet protocol. Beside the Ethernet protocol, the home network also needs a router (which also has an IP address) to allow the communication between the devices (unless there are only 2 devices which we could connect directly). Every home network has a router but also a modem. The modem is need to connect the home network to the ISP: our small LAN becomes then part of the ISP network (formed by all its customers' networks, etc.). The ISP network connects directly to the Internet connecting us to the Internet.
• First, why do we really need the modem and why can we not connect the router in our home network to the ISP network? I understand that a modem is a modulator/demodulator. The modem's input signal looks like a square wave with different discrete voltage levels. The modem's output signal is different: the input discrete signal modulates an electrical sinusoidal carrier signal using either FSK, ASK modulation. Why can we not connect to the ISP without using a modem? Why is the modulation necessary? Without modulation, would the input signal lose too much energy while it is transmitted to the ISP premises? I think so. Using a high frequency electrical signal as a carrier prevents loses...Is that correct?
• What would be needed to be able to connect to the internet directly, without engaging the ISP? What kind of specialized equipment does an ISP have? I am trying to envision the type of hardware that is involved...

Thanks!
Fog37

pbuk
Gold Member
• First, why do we really need the modem and why can we not connect the router in our home network to the ISP network? I understand that a modem is a modulator/demodulator. The modem's input signal looks like a square wave with different discrete voltage levels. The modem's output signal is different: the input discrete signal modulates an electrical sinusoidal carrier signal using either FSK, ASK modulation. Why can we not connect to the ISP without using a modem? Why is the modulation necessary? Without modulation, would the input signal lose too much energy while it is transmitted to the ISP premises? I think so. Using a high frequency electrical signal as a carrier prevents loses...Is that correct?
The situation you describe has not been the case for about 20 years in most countries. The word 'modem' has become a misnomer because there is no analogue signal that is modulated, the 'modem' simply converts between digital ethernet signals on your LAN (which have a limited transmission range) and the digital signals used by the telecoms provider (which have a much greater range).

https://en.wikipedia.org/wiki/DSL_modem
https://en.wikipedia.org/wiki/Cable_modem

• What would be needed to be able to connect to the internet directly, without engaging the ISP? What kind of specialized equipment does an ISP have? I am trying to envision the type of hardware that is involved...

https://en.wikipedia.org/wiki/Tier_1_network

russ_watters and fog37
The situation you describe has not been the case for about 20 years in most countries. The word 'modem' has become a misnomer because there is no analogue signal that is modulated, the 'modem' simply converts between digital ethernet signals on your LAN (which have a limited transmission range) and the digital signals used by the telecoms provider (which have a much greater range).

https://en.wikipedia.org/wiki/DSL_modem
https://en.wikipedia.org/wiki/Cable_modem

https://en.wikipedia.org/wiki/Tier_1_network

Thank you pbuk.

I will read up. I guess ethernet is a protocol for smaller network. It can be applied to different physical media, i.e. we can have ethernet using copper wires, Wi-fi, fiber optics, etc.

Thanks again

rcgldr
Homework Helper
If your ISP and modem support multiple IP addresses, then your home network is essentially an extension of the WAN, and each device connect to your home network gets it's own WAN IP address. In this case a router is not needed, and a switcher can be used instead to connect multiple devices to the modem. However, if one of the devices connected to the modem is a router, then you'd have a LAN on the local side of the router, but still a WAN on the modem.

At one time, my ISP allowed up to 3 separate IP's on a single cable modem. I don't know what the limit is for cable modems. The DSL ISP in my area has an option for up to 5 separate IP's (actually 8 IP's, but 3 of those are used by the ISP).

The situation you describe has not been the case for about 20 years in most countries. The word 'modem' has become a misnomer because there is no analogue signal that is modulated, the 'modem' simply converts between digital ethernet signals on your LAN (which have a limited transmission range) and the digital signals used by the telecoms provider (which have a much greater range).

https://en.wikipedia.org/wiki/DSL_modem
https://en.wikipedia.org/wiki/Cable_modem
It's indeed usual to call the signal on DSL line, or the signal of a cable modem digital, but what happens is really the same as what a modem for the telephone system does. A DSL modem uses a form of frequency division multiplexing, A cable modem modulates to a very high frequency. (There are 20 frequency bands between 500 Mhz and 1500 Mhz).

rbelli1 and pbuk
An ISP will have something called a "Point of Presense" on the internet. This will include a router running BGP (Border Gateway Protocol) which communicates with all other BGP routers on the internet to determine paths to each network. BGP is the backbone of the internet. It's a pretty lame protocol with no security built that auto replicates it's routes so if you make a mistake on a route you can mess up routing for a large part of the internet. You can also re-route traffic which China has done "accidentally" to re-route US traffic through China.

The term modem we use to refer to the old "dial up" connections via a regular phone line in the earlier days of the internet that made those funky noises while dialing up. It's just a means to establish a routed network connection to the ISP who then routes you through their Point of Presense and onto the wider internet. This is all you really need, a routed connection to the internet.

Ultimately the internet as you say is just a large network made up of many smaller networks with routing involved so traffic can move between them. There is nothing "special" per se about it apart from it's size. Moore's Law plays a part here also, the value of a network is proportional to the numbers of nodes connected to it. What use would your computer be these days if it were not on the internet.

What are you trying to achieve by bypassing the ISP? I'm sure there are easier ways to accomplish whatever it is you are trying to accomplish without having to shell out tens of thousands of pounds to establish a Point of Presense on the internet. I'm sure we can offer suggestion to assist if you let us know what you are trying to do.

fog37 and anorlunda
An ISP will have something called a "Point of Presense" on the internet. This will include a router running BGP (Border Gateway Protocol) which communicates with all other BGP routers on the internet to determine paths to each network. BGP is the backbone of the internet. It's a pretty lame protocol with no security built that auto replicates it's routes so if you make a mistake on a route you can mess up routing for a large part of the internet. You can also re-route traffic which China has done "accidentally" to re-route US traffic through China.

The term modem we use to refer to the old "dial up" connections via a regular phone line in the earlier days of the internet that made those funky noises while dialing up. It's just a means to establish a routed network connection to the ISP who then routes you through their Point of Presense and onto the wider internet. This is all you really need, a routed connection to the internet.

Ultimately the internet as you say is just a large network made up of many smaller networks with routing involved so traffic can move between them. There is nothing "special" per se about it apart from it's size. Moore's Law plays a part here also, the value of a network is proportional to the numbers of nodes connected to it. What use would your computer be these days if it were not on the internet.

What are you trying to achieve by bypassing the ISP? I'm sure there are easier ways to accomplish whatever it is you are trying to accomplish without having to shell out tens of thousands of pounds to establish a Point of Presense on the internet. I'm sure we can offer suggestion to assist if you let us know what you are trying to do.
Thanks MikeeMiracle.

Not trying to bypass the ISP but just interested in better understanding how my small home network works and how it relates to the ISP network, the internet, etc.

It may be true that today a "modem" does generate analog signals but, in the case of ADSL, the electrical signal is a carrier of a certain frequency that is carrying a digital signal via a modulation scheme like ASK, FSK, etc.

My small network has internal (private) IP address for each machine connected to the router. The router or gateway (the modem) has a single public IP address. So all the ISP's customer like me have a public (external) IP address that other machines on the internet use to communicate with...

Without the ISP providing my small home network internet service, I would not be able to use the internet and connect to other nodes on the internet. That is why I asked the question. I was just interested in understanding what and how the ISP does that...

Tom.G
Your question is related to networking, the internet just seems to be the final destination.

It's is also highly unlikely that you are using a publicly available IP address, as in anyone connected to the internet can route to it. These are only needed if you are hosting internet resources like a web or e-mail server. They are in short supply and you normally have to pay extra to have one. Unless your hosting some services it's a waste and also a security risk by allowing the general internet to connect directly to your home network.

pbuk
Gold Member
It's is also highly unlikely that you are using a publicly available IP address, as in anyone connected to the internet can route to it.
Not correct.

These are only needed if you are hosting internet resources like a web or e-mail server.
Or expecting a response from a web server, or joining in a Zoom session, or doing pretty much anything else on the internet.

It may well be the case that you do not have a static IP address i.e. a fixed one permanently allocated to you, but if you are connected to the internet then you will have a public IP address. Note that some vendors in some countries may not actually sell you a connection to the internet, instead you are connected to their private network and can only access internet services through their proxy servers, but apart from AOL this has not generally been the case for about 20 years.

Disagree, where your traffic hits the internet, which is usually at the ISP, that device has to have a public IP. Everything behind that can be in private IP ranges as it's all internal routing. This is nothing to do with using Proxy servers.

When you initiate a request for a web page it's your computer reaching out to the web server to establish a connection and transfer data, not the other way around. That web server can't arbitrarily decide to connect to your computer and transfer data as it has no path to do so without your computer reaching out to establish that connection first.

pbuk
Gold Member
Disagree, where your traffic hits the internet, which is usually at the ISP, that device has to have a public IP. Everything behind that can be in private IP ranges as it's all internal routing.
I see you are in the UK, can you give an example of an ISP which does not provide public IP addresses to subscribers?

When you initiate a request for a web page it's your computer reaching out to the web server to establish a connection and transfer data, not the other way around.
Yes, and when it does so your router attaches its public IP address so that the response can find its way back to you.

That web server can't arbitrarily decide to connect to your computer and transfer data as it has no path to do so without your computer reaching out to establish that connection first.
Most routers have the ability to forward incoming requests to a device connected to the LAN: this is pretty much essential for hardcore console gaming.

pbuk
Gold Member
1. visit a site such as whatismyipaddress.com or simply type "what is my ip" into google.

If the two are the same then you have a public IP address. If they are different then your ISP is intercepting your request to whatismyip, adding its own public IP address (which is what you are seeing from whatismyip) and then forwarding the response on to your router.

fog37
No "residential" internet connections have their own dedicated public IP address's in the UK unless you ask for one as far as I know, some don't even offer it as an option. I'm using a "business" ISP and I still had to pay to have one so I could run my own e-mail server.

Are you saying in the US everyone gets their own public routable IP address?

I can't quote for some reason on this site, but in response to the comment about the web server finding it's way back to you. That would respond to whatever IP the external public facing ISP router which initiated the connection has and internal routing within the ISP would then route the connection back to you.

The external WAN interface of your home router may not be in the "standard" private IP ranges but they wouldn't be exposed to the internet as such directly. I guess things work differently in the US.

fog37
pbuk
Gold Member
I am in the UK. My current home internet connection (like every other home internet connection I have had or known) has what you refer to as a 'publicly routable IP address' which is allocated dynamically (by DHCP). The only difference between this dynamic IP address and the static IP address that you pay for is that if I reboot my modem it might connect to a different address when it boots up (in practice, it doesn't).

fog37
jtbell
Mentor
I'm in the US, using a residential DSL internet service from my landline phone company.

I just now used http://whatismyipaddress.com/ to find out my public IP address, then shut down my DSL modem, restarted it, and checked my public IP address again. The two addresses were different in the lowest-order portion: old = 13, new = 101.

fog37
That has been my understanding: all the devices in my small home network (LAN) have their own individual private, internal IP which was assigned to them by the router. All devices, jointly all have the same and one public and dynamic IP which was assigned by the ISP to my network. I believe that the public and dynamic IP is the IP of my router?! It costs lots of $$to have a static public IP, as far as I understand. Going back to my router: today, an ISP provides a box that is both a wired and wireless router and a gateway, all in one. The term gateway is synonym for "modem". Using this "router/modem", my home network becomes connected to the LAN of the ISP. So I would say that my home network is a sub-network of the ISP network (which is a LAN). At a cost, we can now have fiber optics cabling connected directly to our home premises. That is super fast. So I guess the bottleneck is then the ethernet cables or the wifi signal strength at my house. However, if we use an ethernet cable that is short and we connect directly to the gateway box, we should be able to experience fiber optics fast speeds. At the house premises there is some optical to electrical signal conversion technology converting the high speed light signal running through the fiber into an ethernet signal.... Thanks for all your contributions on this topic. anorlunda Staff Emeritus I think this discussion mixes the meanings of public and static or public and permanent. Even a dynamic IP address is public, but it is not permanent or static. You could run a web server on your PC and test your web page, using a second computer (or the same computer) as long as you know the IP address to type into the tester's browser. But if you want to address the web page with a domain name, then the domain name must be associated with a static IP address. The DNS server gets your request for the domain name, and looks up the corresponding IP address, so it needs the IP address in advance. Try this yourself. From the command prompt, type SYSSTEMINFO. That will show your current IP address. Then type PING and the IP address. It will respond. Then call your brother in another state, and ask him to PING the same IP address. It should work for him also. In that respect, the IP address is public. pbuk Science Advisor Gold Member But if you want to address the web page with a domain name, then the domain name must be associated with a static IP address. The IP address does not need to be static, there are DNS services that specialise in dynamic routing (dyndns, noip...) and many routers support this automatically. The DNS server gets your request for the domain name, and looks up the corresponding IP address, so it needs the IP address in advance. Yes, but it only needs it about 100ms in advance! Try this yourself. From the command prompt, type SYSSTEMINFO. That will show your current IP address. Then type PING and the IP address. It will respond. Then call your brother in another state, and ask him to PING the same IP address. It should work for him also. In that respect, the IP address is public. systeminfo will only show your the LAN address of the device (usually starting 192.168. but it might start 10.) If you want to find the WAN address of your router then you need to check your router, or a site like showmyip. When you ping that IP it is your router that responds, not the device you are using (unless you have set your router to forward ping traffic to your device). fog37 and anorlunda jbriggs444 Science Advisor Homework Helper That has been my understanding: all the devices in my small home network (LAN) have their own individual private, internal IP which was assigned to them by the router. All devices, jointly all have the same and one public and dynamic IP which was assigned by the ISP to my network. I believe that the public and dynamic IP is the IP of my router?! It costs lots of$$ to have a static public IP, as far as I understand.
It turns out that, in most areas, a business class ISP connection is not that expensive. I do not know what our company pays. Just that it is surprisingly inexpensive compared to leased lines.

The standard service that we order for most company locations is an Ethernet handoff with a one public routable Internet IP address. The router which plugs into that Ethernet connection acquires the public IP address using DHCP, just like a typical home router would do.

In some areas IP addresses are tight and the local ISPs may balk at the idea of providing a publicly routable IP address. However, that problem affects only a small fraction (less than five percent) of the locations where my company has connectivity. We run Cisco SDWAN (formerly Viptela). The devices that we plug into the Ethernet handoff can interoperate with each other over the public Internet as long as only one of the two ends sits behind a NAT (network address translation) arrangement.

Going back to my router: today, an ISP provides a box that is both a wired and wireless router and a gateway, all in one. The term gateway is synonym for "modem". Using this "router/modem", my home network becomes connected to the LAN of the ISP. So I would say that my home network is a sub-network of the ISP network (which is a LAN).
The typical arrangement is not what I would call a "sub-network". That term is reserved for the practice of sub-dividing a chunk of allocated address space into smaller pieces and handing out those pieces. Some of old-timers even recall the days when a "subnet" had an even more restrictive meaning within the old classful way of doing business. We do CIDR (RFC 1519) now and have done so for three decades.

Instead, the typical arrangement involves Network Address Translation (NAT). The customer's home LAN runs on address space defined in RFC 1918. The relevant ranges are 192.168.0.0/16, 172.16.0.0/12 and 10.0.0.0/8. The ranges typically seen on home routers are 192.168.0.0/24 or 192.168.1.0/24, but the defaults vary. Customers may have access to their own routers and a corresponding ability to choose a particular RFC 1918 subnet on their own.

Here is how NAT works. Say you have a client on the local home network making an outbound connection to a web site whose IP address turns out to be 1.2.3.4. The web browser will open up a TCP connection uniquely associated with five pieces of information:

Source IP: 192.168.1.100 [on the home LAN]
Destination IP: 1.2.3.4 [somewhere on the public Internet]
Protocol: tcp [IP protocol number 6]
Source port: 56789 [some random high number not currently in use by the client]
Destination port: 443 [assuming you are using a https: site]

When this connection goes outbound through a consumer grade home router (or an appropriately configured professional grade router for that matter) the router will keep track of the connection and will adjust the packet:

Source IP: the router's assigned public IP
Destination IP: 1.2.3.4 [unchanged - we're still connecting to the same server]
Protocol: tcp [still TCP, we're still making a TLS connection]
Source port: 56789 [The router could allocate its own random source port, but most consumer grade routers will leave the source port alone as long as no other connection through the router is already using it. In an ISP environment with double-NAT, port exhaustion is a concern here and good routers may be needed. A home environment is pretty safe from that].
Destination port: 443 [it's still an https: site so we're still using TLS]

As far as the ISP is concerned, your router is doing business from a single IP.
As far as the home network is concerned, it has direct visibility to the public Internet.

When a response packet comes back from the target web server, the router adjusts the packet so that the home network client sees what it expects to see.

The above arrangement for outbound NAT is called different things by different vendors. "Dynamic NAT" or "PAT" are common terms.

When a connection comes inbound through a consumer grade router, life is not so simple. Note that it's been ten years or more since I've messed with a home router for this. What you will often have is the ability to designate, for each TCP port on which your home router might receive traffic, a particular internal IP address. When a packet comes in addressed to port 443, for instance, the router can statically NAT this to the designated internal destination IP address, leaving the rest of the packet fields unchanged. Then when the return traffic goes back out, the static NAT is reverted, leaving the packet with the router IP in the source address field.

In professional circles, this would be "static NAT". I think home routers call these "bypass rules" or something similar.

Not directly relevant to your situation, but in our company's environment, we use Cisco SDWAN routers. They can operate successfully even behind a layer of consumer grade NAT. To accomplish this feat, they make use of an external broker device ("vbond", "vsmart" and "vmanage") to keep a roster of the interoperating public and private IPs and the UDP ports needed to penetrate the relevant NAT devices.

We have a small set of SOHO users who have a vEdge device at their home and obtain direct participation in the SDWAN network through that conduit.

The bulk of our company's users are on Cisco Anyconnect which makes use of TLS or DTLS. Looking over at my work connection, I see that it is currently on DTLS 1.2 using ECDHE_ECDSA_AES256_GCM_SHA384. I have a physical home LAN IP address of 192.168.1.171 and a default gateway of 192.168.1.1. However, the VPN adapter feeding back to the company network is at 10.55.231.179 with a gateway of 10.55.192.1. The VPN route has a better preference, so it wins [meanwhile a host route takes care of getting the encapsulated traffic shipped on the physical connection to the VPN endpoint].

My SOHO router does PAT on the VPN traffic so that what my company VPN device sees is some public IP (that I could look up if I cared to) instead of my 192.168.1.171 address.

If I browsed the internet from my work machine, things get yet another layer more complex as my outbound traffic uses WCCP on a central Internet router after passing outbound through the corporate firewall and traverses one of a set transparent proxy servers and emerges with yet another publicly routable IP address.

Last edited:
pbuk and fog37
rbelli1
Gold Member
It may be true that today a "modem" does generate analog signals but, in the case of ADSL, the electrical signal is a carrier of a certain frequency that is carrying a digital signal via a modulation scheme like ASK, FSK, etc.

If you object to an ADSL modem being called a modem on that basis than a POTS modem would also not be considered a modem. Actually in that case modems don't exist.

BoB

Tom.G
pbuk
Gold Member
What a comprehensively informative post @jbriggs444! I can add a couple of things:

It turns out that, in most areas, a business class ISP connection is not that expensive. I do not know what our company pays. Just that it is surprisingly inexpensive compared to leased lines.
In the UK one provider will add a static IP to your domestic broadband package for a one-off charge of £5; other providers require you to upgrade to a 'business' broadband which is not prohibitive (but can't be directly compared because domestic broadband is normally bundled with telephony and sometimes media services).

In professional circles, this would be "static NAT". I think home routers call these "bypass rules" or something similar.
It's usually called "port forwarding".

jbriggs444
Hello again,

Just to wrap up this thread (let me know if it has already been explained in one of the replies), the "modem" does not convert a signal from digital to analog (it used to). I believe the explanation was that the modem in a home network converts the digital ethernet signal running on the home network to a different type of digital signal that the ISP can work with...What kind of digital signal is that? What protocol does it use?

Thanks!

jbriggs444
Homework Helper
Just to wrap up this thread (let me know if it has already been explained in one of the replies), the "modem" does not convert a signal from digital to analog (it used to). I believe the explanation was that the modem in a home network converts the digital ethernet signal running on the home network to a different type of digital signal that the ISP can work with...What kind of digital signal is that? What protocol does it use?
Take a look at the article on DOCSIS

wiki said:
Physical layer
• Channel width:
• Downstream: All versions of DOCSIS earlier than 3.1 use either 6 MHz channels (e.g. North America) or 8 MHz channels ("EuroDOCSIS"). DOCSIS 3.1 uses channel bandwidths of up to 192 MHz in the downstream.[14]
• Upstream: DOCSIS 1.0/1.1 specifies channel widths between 200 kHz and 3.2 MHz. DOCSIS 2.0 & 3.0 specify 6.4 MHz, but can use the earlier, narrower channel widths for backward compatibility. DOCSIS 3.1 uses channel bandwidths of up to 96 MHz in the upstream.
• Modulation:
• Downstream: All versions of DOCSIS prior to 3.1 specify that 64-level or 256-level QAM (64-QAM or 256-QAM) be used for modulation of downstream data, using the ITU-T J.83-Annex B standard[15] for 6 MHz channel operation, and the DVB-C modulation standard for 8 MHz (EuroDOCSIS) operation. DOCSIS 3.1 adds 16-QAM, 128-QAM, 512-QAM, 1024-QAM, 2048-QAM and 4096-QAM, with optional support of 8192-QAM/16384-QAM.
• Upstream: Upstream data uses QPSK or 16-level QAM (16-QAM) for DOCSIS 1.x, while QPSK, 8-QAM, 16-QAM, 32-QAM, and 64-QAM are used for DOCSIS 2.0 & 3.0. DOCSIS 2.0 & 3.0 also support 128-QAM with trellis coded modulation in S-CDMA mode (with an effective spectral efficiency equivalent to that of 64-QAM). DOCSIS 3.1 supports data modulations from QPSK up to 1024-QAM, with optional support for 2048-QAM and 4096-QAM.
Personally, I am a software type. I know in general terms about phase shift keying, trellis coding and various schemes for multi-access media (thinking back to things like Aloha net). But my hat is off to the folks who do that stuff for a living.

As you can see from the description above, putting a digital Ethernet signal onto a broadband medium does involve modulation of a digital signal onto analog media.

Last edited:
anorlunda
Staff Emeritus
It becomes a matter of definition. In a strict definition of digital, the signal moves between two DC voltages, say 0V to 5V. But that is not suitable for long wires. But two tones with different frequencies are suitable for long distance transmission, hence FSK modulation.

By contrast in power grids, the signals most critical to security are backed up by analog measurements, transmitted long distances (over dedicated pairs of wires) by a V to F (voltage to frequency) converter, and received by a F to V converter. Unlike FSK, this scheme uses a continuum of frequencies, so we say it is analog, not digital.

jbriggs444
rbelli1
Gold Member
In a strict definition of digital, the signal moves between two DC voltages, say 0V to 5V.

Modern copper ethernet (100Mbps and above) uses more than 2 levels so it is not digital either. The signal on a home lan is only digital inside the various boxes unless you are using unusual conections.

BoB

fog37