I have some basic knowledge about protocols and the OSI model. I know, for example, a good deal about how HTTP and FTP works or the differences between TCP and UDP, but I never studied much of the specifics of internet protocols. Still, I have a question about the lower level of the modem (I'm using an ADSL connection).

I understand that the common ADSL modem needs authentication and is able to communicate with a server to retrieve an IP number and connect with other computers through the internet, right?

But how does my modem connect to this "server" the first time? When I set up my internet the only thing I really did was connecting the modem to a landline, configuring some settings and authentication info and voilá.

I'm positive that I never configured a server address and I believe there would be no address; since, this happened before I had an IP address.

How does an ADSL Modem discover the network upon physical setup?

2 Answers 2


Here's how it work with my setup:

The modem will try to negotiate layer one communication with the DSLAM. I don't know much about this part, so I won't go into that.

I have a router (which in general can be integrated into modem) that initiates PPPoE session and the modem then encapsulates PPPoE into PPPoA. Different set-ups do this part of the story differently. In some cases router may speak PPPoA directly or modem may unpack the PPPoE and repack it as PPPoA. In some cases, Ethernet may be used in provider's core network instead of ATM and then there won't be PPPoA at all.

The router then keeps broadcasting PPP over Ethernet Active Discovery Initiation (PADI) packets and waits for something to respond to it.

The packets go through the core network of my telecom and reach a broadband remote access server. In my case, there are several of them. Each of them sends a reply to the router's WAN side MAC address. The type of reply packet is called PPP over Ethernet Active Discovery Offer (PADO) and contains information such as name of the device that sent the reply, service names, unique identification and so on.

After my router receives first of the PADOs, it sends PPP Active Discovery Request.

It then finishes the PPP over Ethernet Active Discovery phase, initiates a session and starts communication using PPP Link Control Protocol. I think that session number is used to maintain connection with BRAS.

Router sends a PPP LCP Configuration Request packet.

BRAS should first send an LCP Configuration Acknowledge packet and then respond with configuration request type of packet and indicates what types of authentication protocols it supports. In my case that's only Password Authentication Protocol (PAP).

Router receives the configuration request from BRAS and sends out acknowledge.

After that, using PAP in an Authenticate Request router sends out username and password (and since PAP is used, it's in plain-text, open for everyone to see).

BRAS then responds with PAP Authenticate Acknowledge packet indicating successful authentication. The authentication itself uses usually another specialized authentication, authorization and accounting server that's connected to BRAS.

After that, they switch to Internet Protocol Control Protocol. In networks such as this, it's used instead of DHCP.

Router sends IPCP Configuration Request and indicates which parameters it would like to get, in my case IP address and primary and secondary DNS servers.

After that, it sends out IPv6CP Configuration request together with a device identifier.

BRAS then sends out IPCP Configuration Acknowledge packet with IPv4 address.

Next, BRAS sends out IPCP Protocol Reject package and rejects PPP IPv6CP Request.

After that, my router sends out a IPCP Configuration request again and receives IPCP Configuration ACknowledge with IPv4 address and DNS server addresses.

After all that, BRAS sends out PPPoE Active Discovery Session-confirmation packet and from that point on, I have Internet connectivity. Communication then goes onfrom me through my telecom's network into my ISP's network and then onto Internet.

One more thing I should mention is that once every second after LCP starts, my router sends out LCP Echo Request and gets LCP Echo Reply. Using this, it can confirm that it does have Internet connectivity.

Finally in order to terminate session, PPP Active Discovery Termination packet can be sent, or a device can just skip enough Echos to have the connection die to to lack of keepalive messages.

In my case, PADT isn't used. Instead router sends out LCP Termination Request and after it receives LCP Termination Acknowledge, it shuts down the WAN interface.


There are two parts to this:

1) The modem connects to the modem on the other end of the line. This involves a process of "training", finding out what bandwidth can be achieved on the line by measuring its electrical properties. At the end of this process, the modems can exchange digital data.

2) Now there is a digital link, the modem will send out a DHCP request. This is a "link local broadcast" - it goes to the other modem and the network it's on, but no further. Your ISP will send a response with an IP address in and the address of the default gateway to use.

  • thanks i had this curiosity for a long time! Which kind of "modem" is there on the other side? does this all happen on the physical layer? is there a protocol that "formalizes" this "modem-to-modem conversation" Feb 5, 2013 at 17:42
  • 1
    At the other end there will be a DSLAM .For the protocol that would be the G.992 set of protocols (look at the list on the right side to see the different versions) Feb 5, 2013 at 20:16

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .