WO1998020649A1

WO1998020649A1 - Network adapter utilizing an ethernet protocol and utilizing a digital subscriber line physical layer driver

Info

Publication number: WO1998020649A1
Application number: PCT/US1997/020592
Authority: WO
Inventors: Russell W. Bell
Original assignee: Advanced Micro Devices, Inc.
Priority date: 1996-11-08
Filing date: 1997-11-07
Publication date: 1998-05-14
Also published as: DE69733075D1; DE69733075T2; US6052380A; EP0937354A1; EP0937354B1

Abstract

An improved network adapter and architecture which combines the Ethernet bus access protocol with a digital subscriber line-based physical layer driver to provide improved performance. The use of a digital subscriber line-based physical layer driver, such as the DSL physical layer, in conjunction with the Ethernet protocol provides low cost and compatibility advantages associated with Ethernet in conjunction with the improved reliability, robustness, bandwidth, and noise resistance of the DSL layer. This allows standard Ethernet networking on noisy transmission media such as category 3 twisted pair, power lines, and other previously unusable channels.

Description

_TITLE: NNEETTWWOORRKK AADDAAPPTTEERR UUTTIILLIIZZIINNGG AANN EETTHHEERRNNEETT PROTOCOL AND UTILIZING A DIGITAL SUBSCRIBER LINE PHYSICAL LAYER DRIVER

BACKGROUND OF THE INVENTION

Field of the Invention

The present invention relates to networking technology, and more particularly to a local area network which utili/cs the Ethernet protocol for accessing the bus. and which utilizes a digital subscriber line physical layer protocol, such as discrete multi-tone (DMT) modulation or carπerlcss amplitude phase (CAP) modulation methods, for improved transfer bandwidth and reliability

Description of the Related Art

Computer Networks

Computer networks facilitate the mtcropcrabil of compute! s\ stems and allow a plurality of users to share various elements such as data, applications programs, peripherals, and communication links to other networks and files A local area network (LAN) is essentially a combination of two or more personal computers or workstations that arc physically and logicalls connected to each other Local area networks can be interconnected to olhcr networks in other pans of a building or in other cities, this ty pe of configuration being commonly referred to as a wide area network or WAN

In its simplest form, a network comprises two or more interconnected computers The two or more interconnected computers may share data and/or applications, as well as one or more peripherals, such as a printer A typical corporate network includes at least one dedicated file server, two or more client computers and various shared peripherals The file server is ty pically not used to run application software but rather is used to sen ice requests from the various client computers and to store files that arc created by application programs executing on the client computers The client computer

ides requests to the file server for necessary files and also uses other resources of the network, such as printers and modems Each computer or peripheral connected to the network is referred to as a "node^" on the network

Each node or computer on a network includes a network interface card, also referred to as a LAN adapter, which functions as an interface between the computer and the network cabling The network interface card moves data to and from random access manors inside the computer and also controls the flosv of data in and out of the nctsvork cabling svstcm The network interface card has a spcciali/cd port that matches the electrical signaling standards used on the cable and the specific type of cable connector

OSI Reference Model

Modern netsvorks arc based on the Open Systems Interconnection (OSI) Reference Model for nctssork compatibility and interoperability The OSI scscn layer model defines functions and protocols that enable a ssidc s'arictv of network hardsvarc and softs rc to ssork together The OSI model includes ses-en la ers which encompass the physical netsvork connection as svell as the nctsvorking software The seven layers of the OSI Reference Model include a physical lavcr. data link la er, netssork layer, transport laser, session layer, presentation las er, and application layer The physical layer is responsible for transmission of data bclsscen tsso points or nodes on the netsvork

The phs sical lavcr is implemented bs the netssork interlace card and handles the phs sical signaling on the netsvork. including connectors, timing voltages and other matters The physical las er controls the speed of data on the line, rules for connecting and disconnecting the lines, rules for full duplex and half duplex transmission, and types of cable connectors Thus the physical laser defines the type of svirc. e g . coaxial cable or tssistcd pair, the possible lengths and connections to the s ire. the signals on the svirc. and the interfaces of the cabling system

The second layer, referred to as data link las er or logical link las'cr. is responsible for organizing the bits of data into packets or frames The data link las er is also responsible for sending these frames, in sequence, to their destination and expecting an acknoss lodgment of sas c aπrval of the data If the data is not tcceivcd correctls at the other end. the data link laser is responsible for retransmitting the corrupted frames of data The data link las'cr may be implemented in cither softsvare or hardsvarc

The laser is responsible foi addressing and routing between subnctssorks The netsvork layer determines the data destination and the route that the data should take through the netsvork The netssork laser is also responsible for reformatting a message into packets and directing it to its proper destination The transport layer handles the control or message data movement across the netsvork The transport layer defines hoss messages arc handled, and in particular the manner in svhich the netssork reacts to packets that become lost or other errors that occur Thus the transport las er ensures end to end. error free delivers' The transport layer also controls, addressing, routing, and altcrnatis c routings

The session layer performs s'anous administrate c tasks and security, including encryption if desired Session laser sottssarc manages the entire communications session bs ss nchronizing data floss bctssccn the nodes The session laser is also responsible lor establishing and terminating a session logging in and out and access / security features

The presentation layer provides for the representation of the data, including the formatting and display of data on the computer screen The application laser is the application graphical user interface (GUI) sshich provides basic netssork scrs ices. such as file management and transfer, mail and peripheral access, such as printing

Netsvork Topologies and Protocols

There arc s aπous types of local area netsvork standards and configurations as atlablc todas One choice in configuring a local area netsvork is the netssork topologs . r c . the physical layout and connection of the cables between the ssorkstations and the file scrs crs There arc csscntialls tsvo different methods for implementing a local area netssork. these being referred to as client/sen cr and pccr-to-pccr netssorks In a typical pcer-to-pccr netsvork. any svorkstation or computer can be both a file server as ss'cll as a client searching for data or programs A peer-to-peer netssork comprises a scries of ssorkstations that are usualls linked together in a daisy-chain fashion, ss cre each svorkstation is designated cither as a client or a combination of client and scrs'cr svhen the netsvork is set up A client/server netsvork is the type of LAN most often used today In this type of LAN. one or more central computers, referred to as file servers, are desrgnatcd as central data storage locations and message handlers of the system The remaining computer ssorkstations arc designated as clients and arc all linked to the file seπ'ers

The topology or physical las out of a local area netsvork refers to the was in sshich nodes, c g ssorkstations. printers, file sers'ers and other devices, arc physically connected to each other The physical topologv can lake one of various forms, including a bus topology, ring topology, star topology or a hvbπd topology In a bus lopologs'. a long cable acts as the data passagcsvay or bus of the various nodes The file server, ssorkstations and other dcs ices, such as printers, modems and faxes, arc attached to the cable at dtfferent locations, and data travels to and from the ssorkstations through the cable In a ring topology, the svorkstations arc connected to each other in a daisy-chain fashion and form a circle or ring Data is transmitted from one node to the next, sshcre each node or ssorkstation examines data being passed along the ring If the data is not destined for that ssorkstation. the data is forsvardcd to the next svorkstation and so on Since data tras cls in only one direction, there is no danger of data collision Hosvcvcr. anv break in the connection of the netssork svill cause the entire netsvork to go dosvn or become inoperable

In a star topology, all of the nodes in the netssork arc connected to a central hub to which all connections arc made The central hub is in the form of a ring lopologs as described above, and each node is connected to the central hub through a bus One ads antagc of a star topology is that if one line becomes broken or disconnected, onlv that node becomes inoperable, and the remaining portion of the netsvork remains operable In addition to these topologies, s arious hybrid topologies exist sshich combine the features of star, ring and bus topologies

In addition to the physical topologies described above, a netssork typically includes a logical topology or data transfer protocol, s hich defines the method of data transfer between the various nodes on the netsvork A data transfer protocol is nccessars' for nctssorks. cspccialls' nclsvorks using a bus topologs . because of the possibility that tsvo nodes or ssorkstations may attempt to transmit data at the same time over the common bus In other svords. a data transfer protocol is necessary in a bus topology to pre ent data collision

One popular netssork protocol is referred to as Ethernet, sshich is used for nctssorks utilizing a bus lopologs' Ethernet technology svas dcs eloped in the 1980s and is used in a large number of nctssorking applications, including a large number of local area nclsvorks (LANs) The Ethernet protocol is IEEE standard 802 3

Ethernet is based on a netsvork protocol referred to as CSMA/CD (Carrier Sense Multiple Access / Collision Detection) The Ethernet protocol is sometimes referred to as the " cocktail parts^" algorithm, svhercby an individual listens and ssaits for a silent moment before beginning to talk If by chance, these tsvo individuals "collide^", I c . thev begin speaking at the same time, each ""backs off and allosvs the other person to continue At some point, one person continues speaking (gains access to the channel) and finishes their conversation In a similar manner, an Ethernet network interface card senses the change in s oltage of the bus or cable before attempting to send a packet of data to its destination If no s oltage disruption is detected, the packet of data is transmitted dosvn the cable tosvard its destination Hosvcvcr. if the netssork interface card senses the presence of data, it svaits a random period of time before again attempting to send its packet of data to its destination

The physical lavcr of Ethernet has been used in a s'aπets of transmission media, including radios, coaxial cables, infrared applications, tsvistcd pair sviπng. and others The Ethernet netsvork protocol is mostly used in a tsvistcd pair wiring or coaxial cable ens lronmcnt The maioπts of Ethernet nodes use Manchester encoding at the physical layer This scheme svorks well m a controlled, losv noise, ens'ironmcnt that can tolerate the inefficiency of the Manchester code (signaling rate is tss icc the data rate) Hosvevcr. in the presence of noise, or other interference, or in transmission cns ironmcnts that has c time s arying characteristics, the Ethernet physical layer mav not has c adequate transmission qualify In noisy ens lronmcnts. it is important to have a robust physical layer transport mechanism

Digital Subscriber Line Tcchnolous - DSL

Digital subscriber line (DSL) technologies are currently being developed to more fulls utilize the existing copper svirc infrastructure The ncsv digital subscriber line technologies include ADSL (Asymmetric Digital Subscriber Line). SDSL (S\ iumctπc Digital Subscriber Line), and HDSL (High bit-rale Digital Subscriber Line) among others ADSL svas developed bs companies that supply technology to telephony service providers ADSL is a transport technology that provides higher bandsvidth scrs'iccs over existing copper "loops^" deployed by the phone companies around the ssorld Due to the v alue of this copper, telephony equipment suppliers have developed the above ncsv digital subscriber line technologies svhich require a large amount of processing or MIPS (million instnictions per second) and svhich take ads'antagc of the rapid pace of semiconductor development in order to pros idc increased bandssidth ADSL uses digital signal processing techniques at each end of the copper loop to provide higher bands idth and thus extend the life and bandsvidth of these copper loops Today^'s technology can pros idc data rates from 1 5 Mbps to in excess of 50 Mbps with widely varying cost and distance points ADSL generally uses one of tsvo modulation schemes, these being carπcrless amplitude-phase (CAP) modulation and discrete multitonc (DMT) modulation DMT modulation is a technology that dis ides the available bandsvidth of a channel into sections Each of these sections is "characterized^" for noise and attenuation After the characteristics of the channel arc known, the transmitters and rcccis crs can compensate for the deficiencies in the transmission path The ^"bands^" of the channel that arc deficient rcceis e less data to transmit, and good bands receive more data to transmit The aggregate effect of this channelization is higher bandsvidth over the channel Typical phone lines arc characterized bs 4kHz of bandsvidth Based on typical signal to noise ratios and Shannon^'s Theorem, the maximum capacits of the channel using conventional techniques is under 100kbps Using ADSL. 1 5 to 6 Mbps is possible using the same channel

CAP modulation is also an out-of-band modulation technique ss hich attempts to equalize the line using digital techniques CAP modulation attempts to learn the channel characteristics and appls the ins'crse of these characteristics on the rcceis e side of the channel The net effect of this equalization is to attempt to obtain ""ideal^" channel response

Therefore, a variety of advanced technologies base been dcs_'elopcd to obtain more bandsvidth from a copper- based communication channel such as tssisted pair sviπng Hossevcr most of the copper-based nctssorks are very telephone centric and are based on a classic point to point netsvork. such as those used in the telephone company^'s local loop circuitry Also, standard telephone cabling is category 3 two ssirc unshielded twisted pair (UTP). svhich is generally considered too noisy for netsvork data traffic Therefore, an improved networking arclutecture system and method is desired svhich allosvs computer netssorking. such as Ethernet netssorking. as s cll as nnpros'ed bandwidth. over noiss' cns'ironments. such as conventional telephone copper wiπng

Home Nctsvorks

PC market characteristics indicate that many home users are purchasing a second computer system Not unlike business users, many of these consumers are not in fas-or of purchasing additional peripherals (printers, scanners, etc ) and svonld prefer to share the peripherals that thev alrcadv oss n Therefore, home users with tsvo or more computers desire a netssorking solution.

One solution for providing netssorking in the home is to install specialized cables in the home, such as category 5 tsvistcd pair or coaxial cable, to pros idc connectivity Hosves'cr. this entails considerable trouble and expense Tsvo other connectivity options for a home netsvork include internal posvcr lines and internal phone lines Hosvcvcr. each of these conncctis'its options uses a noisv communications line ss hich is generally deemed inadequate Radio communication av also be used, although use of radio for this purpose has drasvbacks Therefore, an lmproscd networking solution for a home netsvork is desired svhich can use the existing telephone ssiπng or posvcr line infrastructure for network communications

SUMMARY OF THE INVENTION

The present invention comprises an improved netsvork architecture svhich combines the Ethernet bus access protocol with a digital subscriber line-based physical layer driver to provide lmpro cd performance The use of a digital subscriber line-based phs sical laser driver, such as the ADSL physical lavcr. in conjunction svith the Ethernet protocol pres ides losv cost and compatibility advantages associated s ith

Ethernet in conjunction with the impro ed reliability, robustness, bandssidth. and noise resistance ol the ADSL layer This allosvs standard Ethernet netssorking on noisy transmission media such as category 3 tsvistcd pair, posvcr lines, and other previously unusable channels

The netsvork comprises a plurality of computer systems connected bv a transmission media The transmission media may be any of various types of sviπng or cabling, including standard telephony wiring, I c . category 3 tsvistcd pair, power lines, or other wiring or cabling Each computer system or peripheral coupled to the netsvork includes a netssork adapter or netsvork interface card sshich connects to the netsvork

The netsvork adapter includes an Ethernet controller for implementing at least a portion of the Ethernet protocol The netsvork adapter also includes a digital signal processor (DSP) svhich implements a digital subscriber line-based physical layer driver, such as the ADSL physical layer driver, to transmit data onto the transmission media The ADSL phy sical lavcr driver preferably uses discrete multi-tone (DMT) modulation or carπcrlcss amplitude phase (CAP) modulation The digital signal processor implementing the physical layer driver transmits data onto the transmission media based on channel characteristics of the transmission media for improved performance Thus the netsvork interface card implements the Ethernet protocol, but uses the digital subscriber line- based physical layer driver instead of the normal Ethernet physical layer driver This allosvs improved communications and increased bandsvidth over the Ethernet netssork This also allosvs the use of transmission media svhich are otherwise loo noisy or unreliable for data networking, i c . allosvs the use of existing copper infrastructure without resviπng expense

BRIEF DESCRIPTION OF THE DRAWINGS

A better understanding of the present liis cntion can be obtained when the follosving detailed description of the preferred embodiment is considered in conjunction ss ith the follosving drasvings. in svhich

Figure 1 illustrates a simple network according to one embodiment of the invention.

Figure 2 illustrates a netssork interface card comprised in the computer systems of Figure 1. sv herein the netssork interface card implements the Ethernet protocol and uses a digital subscriber line based physical lavcr dπs cr. such as ADSL discrete mullitonc (DMT) or carπcrlcss amplitude phase (CAP) modulation.

Figure 3 illustrates s irlual point-to-point links in a data netsvork.

Figure 4 illustrates data structures created during initialization of nodes on the netssork; and

Figures 5 - 7 arc flowchart diagrams illustrating start-up and initialization sequences of nodes on a netsvork according to the present invention

DETAILED DESCRIPTION OF THE PREFERRED EMBODIM ENT

Computer Netsvork

Referring nosv to Figure 1. a simple computer netsvork 100 is shosvn The netsvork 100 implements portions of a media access controller (MAC) protocol, such as the Ethernet protocol, and a digital subscriber line-based physical layer dπscr protocol for lmproscd performance This combines the losv cost ads'antagcs and compatibility of the Ethernet protocol with the improved bandsvidth performance of ADSL technology In the preferred embodiment, the netsvork 100 implements portions of the Ethernet protocol and a digital subscriber line-based physical las'cr dπs cr protocol for improved performance Hosses er. the netsvork mas implement other media access controller (MAC) protocols, including a USB MAC protocol, among others As shosvn. the netsvork 100 includes a transmission media 102 The transmission media 102 may be anv of various types The network architecture of the present invention allows the use of existing transmission media, such as existing copper wires, which arc generally too noisy for other ty pes of nctssorks In the preferred embodiment, the transmission media 102 is a copper-based sviπng. such as a tsvo svire tsvistcd pair sviring If the netssork 100 is phvsicalls located in a home, the transmission media 102 may be standard telephone sviπng. 1 c . category 3 tss'o svirc tsvistcd pair copper svirc. sshich is conventionally used for telephony connection In other svords. if the netsvork 100 is physically located in a home, the transmission media 102 may be standard telephone sviπng already installed in the home. Alternatively, the transmission media 102 mav be a standard posvcr line, svhcrein the netsvork uses posvcr line carrier (PLC) techniques The transmission media 102 may be any of various other types of network cabling, such as category- 5 tss isted pair, coaxial cable, or optical fiber, among others

The netsvork 100 includes tsvo or more computers 1 12 and 1 14 Each of the computer systems 1 12 and 1 14 is coupled to the transmission media and is adapted for transmitting data onto the transmission media 102 Each of the computer systems 1 12 and 1 14 includes various standard components, including a central processing unit, one or more buses and memory One or more peripherals, such as printer 120. may be connected to the transmission media 102 as separate nodes on the netsvork and rcceis e data from multiple computers Peripheral 120 may also be connected to a port, such as a parallel port or USB interface, of one of the computer systems

Each of the computers 1 12 and 1 14 includes a netsvork adapter according to the present invention for coupling to the transmission media Peripheral devices 120 sshich directly couple to the transmission media 102 also include a netsvork adapter according to the present invention In the follosving description, it is presumed that peripheral dcs icc 120 is coupled to a port of a computer svstcm. such as computer system 1 12. and thus peripheral device 120 docs not include a separate netsvork adapter

As described further bclosv. the network adapter comprised in the computer sy stems 1 12 and 1 14 implements at least a portion of the Ethernet protocol and implements a digital subscriber line-based physical layer driv er to transmit data onto the transmission media 102 This enables each of the computer systems 1 12 and 1 14 to transmit data onto the transmission media 102 based on channel characteristics of the transmission media for improved performance

As noted abov e, the transmission media 102 mav comprise a noisy transmission line, such as standard telephone wiring (category 3 unshielded t isted pair) or a power line The network of the present invention operates to adapt to noise characteristics of the transmission media to prov idc linpiov cd data communication on the transmission media 102

It is noted that a power line contains a relatively large amount of noise and is unsuitable for most modulation and transmission schemes However, the DSL physical lav er driver has robust characteristics which allosv the power line to be used for network traffic

In an embodiment here the transmission media 102 is a two-w ire tsvistcd pair telephone line, the computer systems 1 12 and 1 14 arc operable to transmit data on the transmission media 102 at substantially the same lime as normal telephony s oicc communications on the transmission media 102 This is due to the out- of-band nature of DSL. sshich operates to modulate data on a portion of the frequcncs band outside of the soicc range Thus, both telephony s oicc and netsvork data communications can occur on the transmission media 102 at the same time This enables the telephone line to not be "busy^", i c . allows the home (or office) user to place and receiv e telephone calls, while using the telephone line for netssork data traffic Thus, the use of a DSL physical lavcr for a homc-arca-netssork allosvs data and s oice to coexist on the same cabling at the same time This allosss telephone cons crsations to occur ss lule data files arc being moved inside the home Network Interface Card

Referring noss to Figure 2 a netssork interface card 200 is shosvn according to the present invention As shosvn. the netssork interface card 200 includes a port 202 for coupling to transmission media The netsvork interface card 200 includes an Ethernet controller 212 for implementing at least a portion ol the Ethernet protocol The Ethernet controller 212 implements a collision sense multiple access (CSMA) protocol for initiating accesses onto the transmission media For more information on the Ethernet protocol, please see ""Computer Nctssorks bs Andrew S Tannenbauni. Prentice Hall. 1 88. w hich is hereby incorporated by reference in its entirety

The netssork interlace card 200 includes one or more buffers 214 for storing data to be transmitted through the port 202 onto the transmission media 102. and for receiving and storing data receis ed from the port 202 from the transmission media 102

The network interface card 200 also includes a digital signal processor (DSP) 222 sshich implements a digital subscriber line-based physical las er driver to transmit / receive data onto the transmission media 102 The digital signal processor 222 transmits data onto the transmission media 102 based on channel characteristics ol the transmission media 102 for lmpros cd performance Thus the netsvork interlace card 200 uses the digital subscriber line-based phs sical las'cr dπs cr instead of the normal Ethernet physical laser dπs cr This allosvs lmpros cd communications and increased bandsvidth os cr the Ethernet netsvork and allosvs use of existing wiring infrastructures

The digital signal processor 222 is coupled to the Ethernet controller 212. and the digital signal processor 222 operates to transmit data onto the transmission media 102 when the Ethernet controller 212 indicates that the transmission media 102 is available Thus the network interface card 200 uses at least a portion of the Ethernet protocol for bus access Hosvcs cr. the network interface card 200 docs not use the Ethernet physical laser dπs er. but rather uses a digital subscriber line-based physical laser dπs cr This provides greater bandssidth and better resistance to noise, thus allowing the use of existing copper ss iπng infrastructures sv hich has c more noiss transmission media

The netsvork adapter 200 further includes memory 204 for storing code and data The memory 204 stores code executable bv the Ethernet controller 212 for enabling the Ethernet controller 212 to implement at least a portion of the Ethernet network protocol The memory 204 also stores code and data used bv the DSP 222 for enabling the DSP 222 to implement the digital subscriber line-based physical layer driver The network adapter 200 includes v arious other logic tv picallv comprised in network interface cards, as is svcll known in the art

As noted abosc. the netsvork interlace card 200 includes DSP 222 which implements a digital subscriber line-based physical las'er driv er to transmit/rcccis'c data onto the transmission media 102 In the preferred embodiment, the logic circuitrs ss hich implements the digital subscriber line-based phs sical laser driver is prefcrabls DSP 222 Hosvcvcr it is noted that ans of s arious tspcs of logic circuitrs or programmable processor mav be used In the present disclosure, the term digital signal processor^' is intended to include ans of various types of discrete digital and/or analog logic, programmable logic, or programmable processor including a general purpose CPU or DSP. among other types of logic The digital subscriber line-based physical laser driver uses a Digital Subscriber Line (DSL) technique, such as ADSL (Asymmetric Digital Subscriber Line) techniques. SDSL (Symmetric Digital Subscriber Line) techniques, and/or HDSL (High bit-rate Digital Subscriber Line) techniques, among others, to achieve increased bandwidth and improved communication on a noisv channel The digital subscriber line-based physical layer driver preferably uses DSL techniques, such as discrete multi-tone (DMT) modulation or carrierless amplitude (CAP) modulation, to transmit data onto the transmission media 1 2 Thus the DSP 222 is preferably programmed to use either DMT or CAP modulation in the physical lavcr dπs er

In the preferred embodiment, the netsvork interface card 200 uses DMT modulation In this embodiment, the digital signal processor 222 operates to partition at least a portion of the available bandsvidth of the transmission media channel 102 into a plurality of sections, ss herein each of these plurality of sections has a characteristic signal to noise ratio (SNR) and/or noise attenuation The digital signal processor 222 is operable to determine the characteristics of each of the plurality of sections and provides data onto the transmission media 102 based on the determined characteristics For more information on the DSL physical layer driver of the prclcrrcd embodiment, please see U S Patent No 5.479.447 titled "Method and Apparatus for Adaptis c. Variable Bandsvidth. High-Speed Data Transmission of a Multicarπcr Signal Over Digital

Subscriber Lines ^' svhich is hcrcbv incorporated bv reference as though fully and completely set forth herein For other information on the DSL phy sical layer driv er, please sec Daniel Minoli. Video Dialtonc Technology^". McGraw-Hill. Inc 1 95. which is hereby incorporated bv reference in its entirety

Home Netsvork

As mentioned abos'c. in one embodiment, the netsvork of the present invention is configured in a home or other location using standard telephone wiring In this embodiment, since most homes have multiple phone outlets, computer systems and peripherals arc connected to telephone outlets in different rooms via the existing telephone lines. This provides the ability to use existing w iring in the home, and thus no additional wiring is required In other words, this solution leverages the huge installed base of telephone wiring and equipment

Since both DMT and CAP based technologies arc based on modulation techniques that arc out-of- band. they can operate concurrentls ss ith telephone conversation svithout creating interference Thus computer systems and peripherals can communicate while telephone conversations arc in progress Also, any protocol can run on top of this DMT or CAP based physical layer once the channels have been initialized In the preferred embodiment, an Ethernet media access controller (MAC) is executed on top of this DMT / CAP physical laser

As mentioned above, most of the existing DMT or CAP based phssical laser drivers arc targeted towards a point-to-point netsvork Hosscvcr. as sho n in Figures 1 or 3. most home area nctssorks arc multi- point nctsvorks Therefore, the present invention includes a novel system and method svhich enables a physical laser driver, svhich is primarily designed for a point-to-point netssork. to operate in a multi-point to multi-point netsvork In the preferred embodiment and in the discussion below, the network uses a DMT- bascd physical layer DMT Background

As mentioned above. DMT modulation is a technology that divides the available bandwidth of a channel into sections Each of these sections is ^""characterized^" for noise and attenuation After the characteristics of the channel are known, the transmitters and receiv ers can compensate for the deficiencies in the transmission path The ^"bands^" of the channel that are deficient receive less data to transmit, and good bands receive more data to transmit The aggregate effect of this channelization is higher band idth over the channel As mentioned above, the DMT algorithm is primarily designed for a point-to-point channel The preferred embodiment of the invention provides an initialization sequence and enhancements to allow for operation over a multipoint network typical in businesses and potentially (he home market For more information on DMT modulation, please sec U S Patent No 5.479.447 referenced abov e

Multi-Point Access Protocol Operation

Referring nosv to Figure 3. a multi-point network is shosvn sshich includes a plurality of nodes. As shown, the netssork can be considered as a series of point-to-point nctssorks The point-to-point nctsvorks or v irtual nctsvorks listed below exist inside the physical network of Figure 3

Figure 3 - Point-To-Point Links RN 1 to RN2 RN2 to RN3 RN3 to RN l RN2 to RN 1 RN3 to RN2 RN 1 to RN3

Each of these "links ^" is analogous to a single pomt-lo-point link used bv the telephone company for connection ol their loop equipment, although the distances between nodes is considerably smaller m a home area network The netssoik of the present invention uses the standard DMT algorithm in conjunction with the multi-point initialization and access protocol described herein to provide multi-point operation of a DMT- bascd (or other) physical layer driver in conjunction w ith a local area network In the preferred embodiment. the network is an Ethernet based network

The multi-point initialization and access protocol of the preferred embodiment operates to map out all of the characteristics of the communications paths between each of the point-to-point links in a multi-point netsvork In the follos ing discussion, the multi-point access protocol is described using the example three node netsvork of Figure 3 outlined abov e The multi-point access protocol is not limited to operation w ith three nodes, but rather is generic and can be extended to operate in multi-node networks w ithout any modifications Once each of the point-to-point paths ^"characteristics^' hav e been established, communications can occur between each of the nodes using the appropriate characteristics for each link Figures 5 - 7 Flowchart Diagrams

Referring now to Figures 5 - 7. flowchart diagrams are shown illustrating operation of the multi-point access initialization protocol for each of the nodes on the network As shosvn in Figure 5, in step 502 each node, or the physical laser driver comprised in each node, referred to as ^"PHY^" in Figure 5. "wakes up after posser-up or just after being connected to the multi-node netsvork The node knows nothing except its own internal address In step 504 the node determines its internal address In step 506 the node reads a Sv nc bit in an internal register to see if it is ^"sy nchronized^" with the netsvork. l c . to sec if the node has alrcads performed an initialization in this netsvork. If this bit is determined to be low in step 508. indicating the node is not synchronized, the node begins a sy nchronization sequence A logical one in the Svnc bit indicates that the node "knows^" the equalization parameters for each virtual point-to-point link in the netsvork for ss hich it has access

If the Sync bit is determined to be loss in step 508. in step 12 the node begins a jamming sequence at a predetermined frequcncs and a predetermined signal pattern In the preferred embodiment, the node uses an alternating one zero pattern ( 101010 ) for 100 bit times transmitted in an 1 kHz modulation of an FM carrier centered around 100 kHz Anv other nodes on the network w ill constantly monitor this jamming signal

Once the other nodes has c delected this lamming signal, thev complete the transmission of the current packets and they enter an initialization mode or initialization sequence, ceasing ans further transmission until all of the virtual point-to-point links has c been initialized The initialization sequence is represented as step 514 and is illustrated in Figure 6 As shosvn m Figure 6. upon the entering of the initialization mode or sequence, in step 602 each node sets a Hag to 0. indicating the initialization sequence has begun In step 604. each node generates a random number and in steps 606 and 608 each node starts tsso counters, rcspcctis cls In step 606 each node starts a count dosvn timer that is loaded ssith the random number generated in step 604 In step 608 each nodes starts a modulo counter that is set to the maximum lime that svould be required for all of the nodes on the network to complete their initialization transmissions Tv picallv the size of the modulo counter is equal to the maximum possible size of the random number

Upon expiration of a node s random number timer, as determined m step 610. the appropriate node begins to transmit a specialized header using the lamming channel in step 612 The header contains RNx (timer value node x) (address \ ) Figure 4 illustrates the data structures that arc transferred, as discussed further below This address is seen by every node on the network and is transmitted on the same frequency as the jamming signal

Each respective mode also monitors the jamming channel for other node timeouts in step 16 This monitoring for other node timeouts in step 616 occurs until the modulo counter expires as determined in step 614 Thus ss hen the random number timer expires on another node, this other node also transmits a signal on the jamming channel that contains its timer v alue and address When each of the other nodes detects transmission on the jamming channel bs another node m step 618 each of the other nodes record this address and timer s aluc in step 620 In this manner, all of the nodes record all of the addresses and timer s allies of the other nodes on the netssork After a node records an address and ti er v alue in step 620. in step 622 the node checks its timer against received transmissions and sets a master bit to 1. if appropriate, I e , if the respectiv e node^'s timer has not yet expired Thus, a respective node checks the timer value of es ery other node and sets a master value to 1 if the timer value of the respective node is greater than the timer values of other nodes receis ed on the jamming channel The master bit set for a respcctis'c node indicates that the respective node has the largest timer value and is thus the last to transmit Thus, after a certain period of time, and before the modulo counter expires, each node on the netsvork has its timer expire, and each respectis c node transmits its timer s alue and address for the remaining nodes to "hear^" and record

As mentioned above, each node also monitors its modulo counter in step 14 After the modulo timer has svrappcd. sshich indicates that there can be no other nodes on the netssork that has c not vet timed out and transmitted their timer values and addresses, then each node on the netsvork is now aware of es'ery other node on the netsvork

After the modulo counter expires in step 614. in step 634 remote node Ids arc established After step 634. the respectis c node determines if it is the master node If the respective node is not the master node in step 630 then operation returns to step 616. and the node continues to monitor the jamming channel for other node time outs If the node is the master node, then channel equalization is performed m step 632 The channel equalization performed in step 632 is described further sv ith respect to Figure 7

The last node to transmit determines that it is the last node w hen the modulo timer wraps At this point, each node knows a timer value and an address for ev ery other node The " last node^" to transmit compares its timer v alue with the timer values that it has "heard ^" on the netssork and it w ill realize that it has the highest value timer and is therefore the last node to transmit After the last timer has expired and the master node has been determined in step 630. (hen channel equalization is performed in step 632

Referring nosv to Figure 7. the preferred embodiment includes a channel equalization method svhich allosvs the DSL physical layer in each of the nodes to knosv the transmission characteristics of each of the paths bctsvccn each of the nodes Figure 7 illustrates the channel equalization step 632 referenced in Figure 6 As shown in Figure 7. the channel equalization is performed or called lor each node in step 702 In the prclcrrcd embodiment, the first node that ^"timed out^" (the Master node) begins the initialization process The channel equalization process preferably comprises using a DMT (discrete multi-tone) initialization sequence to determine the characteristics of the one-way point-to-point channel bctsvccn each node It is noted that other initialization sequences mav be used, as desired

The flowchart of Figure 7 is described below using the example of Figure 3 Here it is assumed that the Master node is RN 1 The RN1 node uses the DMT (or equivalent) initialization sequence between RN 1 and RN2 and RN 1 and RN3 RN 1 stores these channel characteristics m the PHY dcv ice In anv subsequent communications between RN 1 and RN3, the appropriate channel characteristics will be used At this point RN 1 has completed its initialization and channel characterization

After this channel equalization is performed as indicated in step 704 in step 706 the RN 1 node generates a special HAND-OFF frame that is transmitted on the jamming channel The frame has the bit pattern RN2. RN2 (the transmission of RN2^'s address t ice in succession) This frame instructs RN2 that it is time for RN2 to determine (he characteristics of the channels RN2 to RN 1 and RN2 to RN3 RN2 uses the DMT (or equivalent) channel characteristics algorithm and "learns ^' the characteristics of the point to point links betsvecn RN2 and RN 1 and R 2 and RN3 At this point. RN2 has completed the initialization sequence The RN2 node stores these channel characteristics which will be used in the future in any communications belsvcen RN2 and RN 1 or RN3 RN2 now sends out a special frame RN3. RN3 RN3 receis es this frame and begins the sequence described above RN3 determines the characteristics of the links RN3 to RN 1 and RN3 to RN2 RN3 stores these characteristics. It is noted that all jamming channel transmissions labeled as RN.x comprise the frame illustrated in Figure 4 This frame comprises the initial value of the random nuinbci generated by RNx folloss'cd by the internal address of RN.x

If node 3 has not seen an additional transmission bctsvccn its initialization sequence and the expiration of the modulo timer, then node 3 (RN3) is the last node m the chain Node 3 recognizes that it is the last node on the link based on timer values At this point. Node 3 transmits a special pattern on the jamming channel RN 1 RN2 RN3. allowing every other node to recognize that the multi-point link is now initialized and every node is now asvarc of cs cry other node Furthermore, every point-to-point link has been characterized using a DMT based algorithm and high speed communications arc noss ready to begin This instructs node three to transmit the initialization complete sequence of RN 1 RN2 RN3

As mentioned above, the first node s hosc random number timer reaches zero is designated as the Master Node This node recognizes it is the master node as it will not have seen ans transmission from anv other node when its timer expires Therefore, it will hasc the losscst timer salue and this information is transmitted to all other nodes as part of the initialization sequence (along with the real address of this node) For algorithmic purposes, each of the nodes uses the internal suffix \\^" for it^'s designator The linked list shosvn in Figure 4 is used by the initialization softssarc to traverse the netsvork topologs Using the three node netsvork example, this linked list comprises three entries, svhcre each entry represents one of the nodes on the network At each local node, the intelligence can recognize the entrs' in the data structure that is associated with the local node based on the address and the timer s alue This linked list is re-linked with the head node designated to be the local node Subsequent initialization algorithms can simpls tras crsc the list to assist in the location of each of the virtual point-to-point segments Once the end point or null pointer is reached, the local intelligence in each node recognizes that it has completed the initialization necessary to establish the channel characteristics for each of the virtual point -to-point links

After the initialization sequence is performed for all nodes, than all of the nodes can begin to communicate ssith each other using a v ariety of types of media access controllers (MACs). including Ethernet If an Ethernet protocol is used, collisions can occur on the transmission media Since a DSL physical layer driver is being used, the physical lay er may not be asvarc of the local collisions This is because there is not a collision detection mechanism analogous to the DC offset caused m a Manchester encoded physical las er svhich is ty pically used bv Ethernet, w hen two signals collide In other words, if multiple nodes or computers both attempt to transmit data at the same tune, comipted data may result on the line This will normally generate a DC blip in a pure Ethernet netssork. but ss ill not generate a DC blip in a ss stem svhich uses a DSL physical layer

In one embodiment of the invention, each node in the system uses higher layers in the protocol stack to resolve the collisions and retransmit the data that was impacted Also, if the bit error rate (BER) exceeds some threshold, or the upper layers in the protocol stack continue to receive cornipted packets, they can communicate with their appropriate physical lavcr dπv er to begin a "reinitialize^" sequence that resets each of the nodes on the link via the jamming channel Similarly, if two or more of the nodes generate identical random numbers there is a finite possibility that the nodes may attempt to initialize the channels at the same time It is likclv that the results of this channel characterization will be unpredictable and will result in unstable operation The higher layer protocol stacks w ill detect this mis-operation in the form of lost packets, higher bit error rates, etc . and can force the link to rcss nchronizc It is unlikely that multiple nodes w ill select the same random number on iterative attempts at initialization

In another embodiment, the initialization scheme described abos c is modified to use dedicated bands of frequencies bctsvccn each of the point-to-point links This method is similar to the abos'c initialization scheme, but includes an additional step during the token pass of RN.x. RN.x in sshich the hand off of initialization is passed from node x-1 to node \ If an additional field svas added at the time of hand-off. the actual operating frequencies that ssere used during the channel equalization process could be passed to node x Node \ could choose to avoid operating at these frequencies during its initialization process Likcss ise. node x could update this field when it passes the initialization token to node RNx+ l During the initialization sequence, node RN x+ l would initialize assuming that the frequencies transmitted in the USED FREQ field svcrc not available and svonld initialize accordingly This ssould allow multiple simultaneous channels to operate os cr the same netsvork. although, since the total bandsvidth is shared, the speeds ssould be slosver Also, the protocol could be modified to negotiate hosv much total bandsv idth is av ailable ov er the sy stem Thus, m one embodiment, the network interface card 200 operates to allocate different frequency bands for transfers betsvcen different nodes Thus, for data transfers bctsvccn a first computer and a second computer, the DSP 222 allocates a first subset of frequency bands or chips For data transfers between the second computer and a third computer, the DSP 222 allocates a second subset of frequency bands or chips Thus, in this embodiment, the network interface card 200 implements "collisionlcss ^" Ethernet Each of the devices has a dedicated channel This operates to piev cnt data collisions and actually allows network traffic between different nodes simultaneously

It is noted that the data rates defined bv DSL in its current form arc not mandatory for operation m the netsvork ss stcm of the present invention Tradeoffs between data rales, transmission distance, semiconductor cost, posscr consumption, etc . arc all possible, thus improving the cosl-cffcctis cncss for a physical lay er Also, as Ethernet MAC protocols arc inherently half-duplex, there is no need for elaborate echo cancellation for the simultaneously operating reverse channel that is present in standard DSL Hosscs cr. this scheme docs not prohibit the operation of DSL in conjunction with full duplex Ethernet

Therefore the networking technology of the present invention combines the best of \DSL and Ethernet The use of the Ethernet protocol prov ides low cost advantages associated w ith the economics of scale attributed to the huge installed base of Ethernet dcs ices in the PC market The use of the DSL phy sical las'cr dπs'cr pros'idcs increased bandsvidth in noisv environments, and semiconductor process tcchnologs allosvs the combination of this technology to be useful in a s'aπets of applications

Although the system and method of the present invention has been described in connection ith the preferred embodiment, it is not intended to be limited to the specific form set forth herein, but on the contrary. it is intended to cover such alternatis es. modifications, and equivalents, as can be reasonably included svithin the spirit and scope of the invention as defined bv the appended claims

Claims

WHAT IS CLAIMED:

1 A network adapter with improved transmission capabilities, comprising

a port for coupling to a transmission media.

an Ethernet controller coupled to the port for implementing at least a portion of the Ethernet protocol, wherein the Ethernet controller implements a collision sense multiple access (CSMA) protocol for initiating accesses onto the transmission media.

one or more buffers for storing data to be transmitted through the port onto the transmission media, and for receiving and storing data received from the port from the transmission media.

a digital signal processor coupled to the port w hich performs digital subscriber line physical lavcr operations to transmit/receiv e data on the transmission media

2 The network adapter of claim 1. w herein the digital signal processor transmits / receiv es data on the transmission media based on channel characteristics of the transmission media for improved performance

3 The network adapter of claim 1. w herein the digital signal processor operates to perform discrete multi-tone operations to transmit/receive data on the transmission media

4 The network adapter of claim 1 w herein the digital signal processor operates to perform carrierless amplitude phase (CAP) operations to transmit/receive data on the transmission media

5 The network adapter of claim 1 w erein the digital signal processor operates to partition at least a portion of the as ailable bandssidth of the transmission media channel into a plurality of sections s\ herein each of these plurality of sections has a characteristic noise and attenuation, sv herein the digital signal processor is operable to determine the characteristics of each of the plurality of sections and operates to prov ide data onto the transmission media based on said determined characteristics

6 The network adapter of claim I . further comprising

memory for storing code and data, w herein the memory stores code executable bv the Ethernet controller for enabling the Ethernet controller to implement at least a portion of the Ethernet network protocol 7 The network adapter of claim 1. wherein the digital signal processor is coupled to the Ethernet controller, and wherein the digital signal processor operates to transmit/receive data on the transmission media when the Ethernet controller indicates that the transmission media is as ilable

8 A computer ss stem with improv ed network connectivity comprising

a central processing unit.

memory coupled to the CPU for storing code and data

at least one bus.

a network adapter coupled to the bus and adapted for interfacing the computer system to a computer network the network adapter comprising

a port lor coupling to a transmission media

an Ethernet controller coupled to the port for implementing at least a portion of the Ethernet protocol ss herein the Ethernet controller implements a collision sense multiple access (CSMA) protocol for initiating accesses onto the transmission media.

one or more buffers for storing data to be transmitted through the port onto the transmission media, and for rcccis nig and storing data received from the port from the transmission media.

a digital signal processor coupled to the port sshich performs digital subscriber line physical lavcr operations to transinit/rcccisc data on the transmission media

9 The computer svstcm of claim 8. ss herein the digital signal processor comprised in the netsvork adapter transmits/rcccis cs data onto the transmission media based on channel characteristics of the transmission media for impros cd performance

10 The computer svstcm of claim 8. ss herein the digital signal processor comprised in the netsvork adapter operates to pcrlorm discrete multi-tone operations to transmit/receive data on the transmission media

1 1 The computer sy stem of claim 8 ss herein the digital signal processor operates to perform carrierless amplitude phase (CAP) operations to transmit/receive data on the transmission media 12 The computer svstcm of claim 8. herein the digital signal processor comprised in the netssork adapter operates to partition at least a portion of the as ailable bandssidth of the transmission media channel into a plurality of sections, ss herein each of these plurahts of sections has a characteristic noise and attenuation, svherein the digital signal processor is operable to determine the characteristics of each of the plurality of sections and operates to provide data onto the transmission media based on said determined characteristics

1.3 The computer system of claim 8. ss herein the netssork adapter further comprises

memory for storing code and data, w herein the memory stores code executable bs the Ethernet controller for enabling the Ethernet controller to implement at least a portion of the Ethernet netsvork protocol

14 The computer svstcm of claim 8. herein the digital signal processor is coupled to the Ethernet controller, and wherein the digital signal processor operates to transmit/receiv e data onto the transmission media w hen the Ethernet controller indicates that the transmission media is av ailable

15 A network which implements portions of the Ethernet protocol and a digital subscriber line-based driver lavcr protocol for improved performance, comprising

a transmission media.

a first computer system including a central processing unit and memory, sv herein the first computer system is coupled to the transmission media and is adapted for transmitting data onto the transmission media.

a second computer sy stem including a central processing unit and memory , sv herein the second computer svstcm is coupled to the tiansnussion media and is adapted for transmitting data onto the transmission media.

ss herein each of the first and second computer sy stems includes a netsvork adapter for coupling to the transmission media, the netssork adapter comprising

a port for coupling to a transmission media

an Ethernet controller coupled to the port for implementing at least a portion of the Ethernet protocol svherein the Ethernet controller implements a collision sense multiple access (CSMA) protocol for initiating accesses onto the transmission media. one or more buffers for storing data to be transmitted through the port onto the transmission media, and for receiving and storing data received from the port from the transmission media.

a digital signal processor coupled to the port svhich performs digital subscriber line physical layer operations to transmit/rcceis'c data on the transmission media

16 The netssork of claim 15. svherein the transmission media comprises an internal posvcr line, svherein the posvcr line transmission media includes noise, svherein the digital signal processor comprised in the netsvork adapter of each of the first and second computer systems operates to adapt to noise characlcristics of the transmission media to provide improv ed data communication on the power line transmission media

17 The network of claim 15, wherein the transmission media is a two-wire twisted pair telephone line, wherein the first and second computer sy stems aic operable to Iransinit/rcccivc voice and data on the transmission media at substantially the same time

18 The network of claim 15. further comprising

one or more peripherals coupled to the transmission media performing input/output functions

19 A netsvork adapter ss ith improved transmission capabilities, comprising

a port for coupling to a transmission media.

an Ethernet controller coupled to the port for implementing at least a portion of the Ethernet protocol.

one or more buffers for storing data to be transmitted through the port onto the transmission media, and for receiving and storing data receis ed from the port from the transmission media.

a digital signal processor coupled to the port sshich performs discrete multi-tone operations to transmit/receive data on the transmission media, svherein the discrete multi-tone operations aic adaptable to channel characteristics of the transmission media for improved performance

20 The netssork adapter of claim 1 . sv herein the digital signal processor operates to partition at least a portion of the as ailable bandsvidth of the transmission media channel into a plurality of sections w herein each of these plurality of sections has a characteristic noise and attenuation, herein the digital signal processor is operable to determine the characteristics of each of the plurality of sections and operates to provide data onto the transmission media based on said determined characteristics 1 A method for providing netsvork data onto a transmission media, comprising

creating one or more packets of data for transmission onto the transmission media for transfer on the network.

monitoring the transmission media to sense the change in s oltage of the transmission media, svherein said monitoring occurs prior to attempting to transmit the one or more packets of data on the transmission media.

svaiting a random period of time before again attempting to transmit the one or more packets of data on the transmission media if said monitoring senses the presence of data on the transmission media: and

transmitting the one or more packets of data on the transmission media if a voltage disruption is not detected, wherein said transmitting uses a subscriber line-based phy sical lavcr driv er ss herein said transmitting the one or more packets ol data is adaptable to channel characteristics of the transmission media for improved performance

22 The method of claim 21. svherein said transmitting the one or more packets of data on the transmission media comprises performing discrete multi-tone operations to transmit data onto the transmission media.

23. The method of claim 21, svherein said transmitting the one or more packets of data on the transmission media comprises performing carrierless amplitude (CAP) operations to transmit data onto the transmission media

24 The method of claim 21. wherein said transmitting the one or more packets of data on the transmission media comprises partitioning at least a portion of the as ilable bandsv idth of the transmission media channel into a plurality of sections, sshcrein each of these plurality of sections has a characteristic noise attenuation, svherein said transmitting determines the characteristics of each of the plurality of sections and provides data onto the transmission media based on said determined characteristics