12.1 Broadband Internet Access

A popular home network application is for sharing of a broadband access service, such as an ADSL line or a cable modem, among multiple PCs and their users in a household. Let us first examine what is involved in connecting a single user to the ADSL service. Figure 12.1 shows the general ADSL installation for a single PC at a household. With the splitterless installation approach, an ADSL modem can be located anywhere in a household as long as a regular RJ-11 telephone jack is available. For this approach, microfilters sometimes might need to be inserted between those RJ-11 jacks and the connected telephone sets to isolate low input impedances of telephone sets at high frequencies from interfering with the proper operation of the ADSL modem. The connection between the ADSL modem and the PC is usually done via a crossover Category 5 patch cable.

Figure 12.1. Broadband Access via ADSL

Let us look at the ADSL end-to-end system architecture to learn what communication protocols are involved in connecting an ADSL user to the Internet [1]. Figure 12.2 shows an ADSL end-to-end architecture from a user to the Internet including ADSL modems, telephone loops (labeled as ADSL), MDF (Main Distribution Frame)/POTS (Plain Old Telephone Set) Splitters, the DSLAM (DSL Access Multiplexer), ATM links including the aggregator, and the ISP. MDF is a physical telephone loop termination hardware platform. POTS splitters separate telephone services from ADSL using a combination of low-pass and high-pass filters. The DSLAM hosts ADSL transceivers at a Central Office (CO). The aggregator does ATM traffic concentration. Sometimes, the name of the NAP (Network Access Provider) is used for the entity providing all necessary equipment and associated services to connect a PC at a household to a router at the ISP. Protocol stacks along the ADSL transmission path are also demonstrated at the bottom part of Figure 12.2. Information in packet format enters the stack from the top and reaches the physical layer at the bottom. Matching physical layers then carry packets from one place to the other. This architecture was also known as the bridged Ethernet over ATM approach. This early approach might have been replaced by the Point-to-Point Protocol (PPP) over ATM architecture where a PPP connection is established between a PC and a router and PPP instead of using Ethernet packets are encapsulated into ATM cells.

Figure 12.2. ADSL End-to-End System Architecture

IP packets that arrive at an ADSL modem in Ethernet format are encapsulated into ATM cells. Within an ATM cell of 53 bytes, 5 bytes are defined as the ATM header, which contains channel and path information directing the cell to its destination. Ethernet packets are actually encapsulated by circuits within the ADSL modem according to rules defined by RFC (Request For Comments) 1483 and AAL5 (ATM Adaptation Layer Type 5). RFC 1483 maps the ATM channel and path as well as payload information into proper parts of a frame to be encapsulated into ATM cells. AAL5 is designated to handle UBR (Unspecified Bit Rate) service, which fits the traffic characteristics of Internet access very well. The encapsulation process, sometimes also called SAR (Segmentation And Reassembly), prepares a cell for transmission in three steps. First, it appends a variable-length PAD and an 8-byte trailer to a packet. The PAD ensures that the resulting frame falls on the 48-byte boundary of an ATM cell. The trailer includes the length of the frame and a 32-bit CRC computed across the entire frame. Second, it segments the frame into 48-byte blocks. Finally, it places each block into the Payload field of an ATM cell. A PVC packet transmission connection is established between the ADSL modem and the ISP router by assigning proper VCI (Virtual Channel Identifier) and a VPI (Virtual Path Identifier) in ATM cell headers via the management system in the initial ADSL installation process. Notice that a DSLAM needs to deal with ADSL traffic only at the ATM cell level. Sometimes, packets are recovered before reaching an ISP router if the ATM connection is not available for that last link.

The initial connection between a PC and an ISP router is established through a few steps involving networking protocols at different levels. First, let us look at the single PC without a home router case. When a PC is turned on, its Ethernet NIC is activated. Ethernet transceivers in the ADSL modem and PC exchange signals and establish a physical connection. Logic Link Control sublayers above both transceivers then identify each other by exchanging packets. Packet exchange between LLC sublayers is capable of MAC address identification via the Exchange Identification (XID) command. Acting as a bridge, the ADSL modem forwards all packets from the PC to the ISP router and relays packets from the router back to the PC. The identity of the user can be confirmed by comparing the PC MAC address with what is provided to the router earlier. After the confirmation of identity, the router can exchange further information with the PC using SNAP, which can deliver IP packets over the LLC sublayer.

The ISP usually assigns a single IP address for each subscriber. Multiple IP addresses are usually available at extra costs. However, this might not be necessary if a home router can be used instead. The ISP can assign either a fixed or a dynamic IP address for its subscriber. The fixed IP address is assigned and reserved whenever a subscriber signs up for the service. A dynamic IP address, on the other hand, can be assigned when a subscriber accesses the Internet, and the same IP address can be reassigned to another subscriber when it is no longer utilized. Because the available number of IP addresses is limited, most ISPs prefer assigning dynamic IP addresses on demand. An IP address is used by Internet applications such as a Web browser. When hot links of a Web page are activated through clicking, IP packets are exchanged between a PC and Web servers across the Internet. The choice of using a fixed or a dynamic IP address also needs to be specified in the network portion of a PC operating system. For Windows operating systems, it is specified by defining TCP/IP (Transmission Control Protocol/Internet Protocol) properties of the corresponding NIC, which can be located through the Network icon within the Control Panel. A fixed IP address and associated subnet mask can be entered or a dynamic IP address can be obtained automatically when the PC is turned on and Internet applications are activated by selecting the corresponding choice.

Along with the assigned fixed IP address, other information such as router (or Gateway) IP address, DNS (Domain Name Server) IP address also need to be specified in the network portion of a PC for Internet applications to function properly. For the dynamically assigned IP address case, DHCP (Dynamic Host Configuration Protocol) is used to obtain an IP address through the ISP router. The PC first sends a broadcast request (called a DISCOVER or DHCPDISCOVER), looking for a DHCP server to answer. The ISP router then directs the DISCOVER packet to the DHCP server. The DHCP server then temporarily reserves an IP address for the PC and sends back an OFFER (or DHCPOFFER) packet, with that address information. The PC sends a REQUEST (or DHCPREQUEST) packet, letting the server know that it intends to use the address. The server sends an ACK (or DHCPACK) packet, confirming that the PC has been given a lease on the address for a specified period of time. The server also configures the PC's DNS servers, WINS (Windows Internet Naming Service) servers, and sometimes other services as well.

Next let us examine what is involved in connecting a single user to the cable modem service. Figure 12.3 shows the general cable modem installation for a single PC in a household. A cable modem can be located anywhere in a household as long as a regular cable TV (RG6) plug is available. The connection between the cable modem and the PC is also via a crossover Category 5 patch cable.

Figure 12.3. Broadband Access via Cable Modem

Let us look at the cable modem end-to-end system architecture to learn what communication protocols are involved in connecting a cable modem user to the Internet [2]. Figure 12.4 shows a cable modem end-to-end architecture from a user to the Internet including cable modems, coaxial feeder and distribution cables, fiber nodes, fiber cables, headend CMTS (Cable Modem Termination Systems), ATM or packet communication links, and the ISP. The CMTS contains a downstream modulator, an upstream demodulator, and backbone network interface adaptors. The headend equipment also provides channel-combining and -splitting functions because modulated cable modem RF signals share the transmission media and frequency spectrum with other cable TV video channels. Protocol stacks along the cable modem transmission path are also demonstrated at the bottom part of Figure 12.4. Not shown is another MAC layer within the HFC (Hybrid Fiber Coaxial) transmission system between cable modems and CMTS. Details of this architecture are defined by specifications under the name of DOCSIS (Data Over Cable System Interface Specification) released by Cable Labs [3].

Figure 12.4. Cable Modem End-to-End System Architecture

Ethernet packets that arrive at a cable modem from a PC are attached with a new MAC header of 6 bytes to form a cable modem MAC frame, scrambled using a generator polynomial of G(x) = x¹⁵ + x¹⁴ + 1, Reed-Solomon encoded with t = 1 10 as specified by CMTS, appended with a preamble of a variable length also specified by CMTS, and modulated to an RF carrier of between 5 and 30 MHz in the upstream direction (from a cable modem to CMTS). Depending on the modulation baud rate, an upstream RF carrier has a bandwidth of between 200 kHz to 3.2 MHz including guard bands. A multiple number of upstream RF carriers can be arranged in the available upstream spectrum of between 5 and 30 MHz. Each upstream RF carrier is shared by a group of cable modems according to the definition of minislots of duration kt, where k = 1, 2, 4, 8, 16, 32, 64 or 128 and t = 6.25 µs. The cable modem MAC is designed to ensure multiple access to a specific upstream RF carrier by multiple cable modems without much collision. Motion Picture Expert Group (MPEG) packets that arrive at a cable modem from a CMTS have a header of 4 bytes and a payload of 184 bytes for a total length of 188 bytes. Each MPEG packet can contain more than one cable modem MAC frame or a partial of a MAC frame.

Both cable modem and CMTS can be LLC hosts and IP hosts. A cable modem and its CMTS can identify each other via the LLC layer through exchange XID packets. IP packets can also be exchanged between a cable modem and CMTS via the LLC layer using either DIX (DEC, Intel, and Xerox also called Ethernet II) or SNAP protocol. Figure 12.5 shows the general Ethernet LLC frame structure [4]. An LLC frame consists of a 2-byte Destination Service Access Point (DSAP) address, a 2-byte Source Service Access Point (SSAP) address, and a 1- or 2-byte Control field and is followed by data of a variable number of bytes. Figure 12.6 shows the Ethernet II frame structure.

Figure 12.5. Ethernet LLC Frame Structure

Figure 12.6. A DIX or Ethernet II Frame

The connection between a cable modem and CMTS can thus be established and maintained. When a PC is turned on, its Ethernet NIC exchanges signals with that of a cable modem to establish a physical connection. A cable modem is capable of acquiring a PC's MAC address manually or through learning. Acting as a bridge, a cable modem forwards packets from the PC to the ISP router and relays packets from the router back to the PC. With limited MAC bridge function, a cable modem can filter out certain packets.