United States Patent
`Allan et al.
`
`19
`
`54 MECHANISM FOR MULTIPLEXING ATM
`AAL5 VIRTUAL CIRCUITS OVER
`ETHERNET
`
`75 Inventors: David Ian Allan; Liam M. Casey, both
`of Ottawa; Andre J. Robert,
`Woodlawn, all of Canada
`73 Assignee: Northern Telecom Limited, Montreal,
`Canada
`
`21 Appl. No.: 08/821,145
`22 Filed:
`Mar 20, 1997
`(51) Int. Cl. ............................................... H04Q 11/04
`52 U.S. Cl. ............................................. 370/397; 370/401
`58 Field of Search ..................................... 370/397, 395,
`370/398, 401, 471, 473, 474
`
`56)
`
`References Cited
`
`U.S. PATENT DOCUMENTS
`
`5,457,681 10/1995 Gaddis et al. ............................ 370/56
`5,490,140 2/1996 Abensour et al.
`... 370/397
`5,490,141 2/1996 Lai et al. ................................ 370/397
`5,732,071 3/1998 Saito et al. .............................. 370/401
`FOREIGN PATENT DOCUMENTS
`
`PCT/CA95/
`OOO29 1/1995 WIPO ............................ HO4L 12/46
`WO95/20282 7/1995 WIPO ............................ HO4L 12/46
`OTHER PUBLICATIONS
`Castro, R. et al., “Support of Data Communications in an
`ATM LAN”, Information Networks and Data Communica
`tion, Proceedings of the IFIP TC6 Int. Conference on
`Information Networks and Data Communication, Madeira
`
`USOO5946313A
`Patent Number:
`Date of Patent:
`
`11
`45)
`
`5,946,313
`Aug. 31, 1999
`
`Apr. 18–21, 1994, pp. 277–295,
`
`Island, Portugal,
`XPOOO5932.98.
`Smith, J.C., “10MBPS to 155MBPS ATM Using the IDT
`SARAMTM as a Concentrator Brouter Core', Wescon 94,
`Western Electronic Show and Convention, Anaheim, Sep.
`27–29, 1994, pp. 495-502, XP000532615.
`Mirchandani, V. et al., “An Internetworking Architecture for
`Integrated Voice and Data Communications”, Proceedings
`of the regional 10 Annual International Conference, Sin
`gapore, Aug. 22-26, 1994, vol. 2, Conf. 9, pp. 749-753,
`XPOOO528218.
`Camarda, P. et al., “A Router for the Interconnection of
`Ethernet Local Area Networks Via an ATM Network',
`Proceedings of the International Conference on Integrated
`Broadband Services and Networks, London, Oct. 15-18,
`1990, pp. 283–288, XPO00410619.
`Paone, R. et al., “Feasibility Model of a Flexible Business
`Customer Premises Network', Electronics and Communi
`cation Engineering Journal, Vol. 4, No. 1, Aug. 1, 1992, pp.
`215-224, XP000307899.
`Primary Examiner-Chi H. Pham
`ASSistant Examiner Maikhanh Tran
`Attorney, Agent, or Firm Aprilia U. Diaconescu
`57
`ABSTRACT
`The invention provides for a E-Mux and a method for
`encapsulating/segmenting ATM cells into/from an Ethernet
`frame at the boundary between an ATM and an Ethernet
`network. An Ethernet end-station on the E-Mux is addressed
`using multiple MAC level identifiers, which are dynamically
`assigned according to the ATM Virtual circuits which ter
`minate on that end Station, and have only transitory signifi
`cance on the Ethernet. A unique ATM OUI identifies the
`frames carrying ATM-traffic.
`
`12 Claims, 10 Drawing Sheets
`
`110
`
`130
`
`140
`
`50
`
`170
`
`Establish Connection
`ATM net-E-Mux
`
`100
`
`120
`
`Receive Cells
`From ATM Net
`
`Mux Cells Payload
`Into PDU payload
`
`Drop Cells
`
`Prepare Frame
`PDU Payload
`Into Frame Payload
`
`160
`
`E-Mux Address
`Into Source MAC
`
`WPFWCInto
`Destination MAC
`
`180N
`Transmit Frame
`over Ethernet
`
`ES 39Accepts Frame
`ES39, 39' Drops Framc
`
`190
`
`200
`ES 39 prepares frame
`Destination ATMOUI/
`Source WPITWC
`
`210
`
`ES 39
`Broadcast Frame
`
`Dest MACc
`ATMOU
`
`YES
`
`
`
`230
`
`LAN traffic
`
`240
`
`
`
`ES39, 39'
`Discard Frame
`
`E-Mux Receive
`Frame From LAN
`
`Pawload
`F
`250
`rame Payloa
`N- Into PDU paylaod
`260
`
`
`
`270
`
`DemuxPDU Payload
`Into Cell Payload
`
`WPI/VC in
`Cells Headers
`
`280
`
`Transmit Cells
`over ATM Network
`
`Comcast, Ex. 1250
`
`1
`
`

`

`U.S. Patent
`US. Patent
`
`Aug. 31, 1999
`Aug. 31, 1999
`
`Sheet 1 of 10
`Sheet 1 0f 10
`
`5,946,313
`5,946,313
`
`5-39.98..E.or
`
`mF
`
`839%?
`
`,§E_
`
`llggllzgwrtue
`
`
`
`<me30E
`
`€<x§av
`
`.m...m..4..m._%ms-l.
`...................mall-l
`........_$4..|.N|N8063.;
`– ?• ?...……“\02
`
`
`
`xc:9mm
`
`
`
`2
`
`
`
`
`

`

`U.S. Patent
`
`Aug. 31, 1999
`
`Sheet 2 of 10
`
`5,946,313
`
`
`
`99
`
`
`
`
`
`
`
`(?V JOJd) g | EHnÐIH
`
`
`
`(uV dolud) OL EHn?IH
`2908
`
`
`
`82
`
`3
`
`

`

`US. Patent
`
`Aug. 31, 1999
`
`Sheet 3 0f 10
`
`5,946,313
`
`AN
`
`_23
`
`mm
`
`-mEomSO
`
`III|||I|V
`mEEooc.
`
`mam_.
`
`58m:
`
`ommmooi
`
`NMEDGE
`
`mm5
`
`mv
`
`8885
`
`8062mm
`
`Du
`
`
`
`52as."m<m30m952“.m
`
`5w2x22-m_
`
`n:
`
`..........................................
`
`n:30umEa"mam
`
`mwmv..Fv-
`
`u:soEEE
`
`Tam
`
`mam
`
`.......................................................................
`
`=9»..EV
`
`4
`
`
`
`
`

`

`US. Patent
`
`Aug. 31, 1999
`
`Sheet 4 0f 10
`
`5,946,313
`
`#N
`
`m;
`
`NIV8:9...
`
`mn_<<
`
`
`
`umoamn._\_n_>
`
`mm
`
`853$
`
`85meE
`
`<mHEDGE
`
`mn_<<
`
`263mm
`
`.<Zw\04._
`
`.881
`
`0..
`
`gNV3.
`
`mg
`
`
`
`032850mo<s_53559
`
`
`
`
`
`.22_o>\_n_>.22
`
`wm
`
`mmon8
`
`5
`
`
`
`
`
`

`

`U.S. Patent
`
`Aug. 31, 1999
`
`Sheet 5 of 10
`
`5,946,313
`
`99
`
`Z– ?
`
`r 8 |
`
`| '-—1–
`
`
`
`
`
`OVW 90/nOSOVW "¡S0C]
`
`
`
`6
`
`

`

`U.S. Patent
`
`Aug. 31, 1999
`
`Sheet 6 of 10
`
`5,946,313
`
`FIGURE 4A
`
`Establish Connection
`ATM net-E-Mux
`
`100
`
`110
`
`130
`
`140
`
`O
`15
`
`170
`
`Saddress
`=VPI/VCI
`
`Receive Cells
`From ATM Net
`
`Mux Cells Payload
`Into PDU payload
`
`YES
`Prepare Frame
`PDU Payload
`Into Frame Payload
`
`
`
`
`
`E-Mux Address
`Into Source MAC
`
`
`
`120
`
`NO
`
`Drop Cells
`
`160
`
`VPI/VC Into
`Destination MAC
`
`180TN Transmit Frame
`Over Ethernet
`
`ES 39Accepts Frame
`ES39, 39' Drops Frame
`
`190
`
`7
`
`

`

`U.S. Patent
`
`Aug. 31, 1999
`
`Sheet 7 of 10
`
`5,946,313
`
`FIGURE 4B
`
`200
`ES 39 prepares frame
`Destination ATMOUIl-1
`Source VPI/VCI
`
`
`
`ES 39
`Broadcast Frame
`
`210
`
`220
`
`
`
`
`
`Dest MAC=
`ATM OUI
`
`NO
`
`LAN traffic
`
`230
`
`YES
`
`
`
`
`
`ES 39', 39'
`Discard Frame
`
`
`
`240
`
`E-Mux Receive
`Frame From LAN
`
`250
`N Frame Payload
`Into PDU paylaod
`
`260
`
`270
`
`DemuxPDU Payload
`Into Cell Payload
`
`VPI/VCI in
`Cells Headers
`
`280
`
`Transmit Cells
`Over ATM Network
`
`8
`
`

`

`US. Patent
`
`Aug. 31, 1999
`
`Sheet 8 0f 10
`
`5,946,313
`
`NN
`
`.23
`
`mm
`
`.mEomSO
`
`mEEooc.
`
`mamB
`
`.8me
`
`8885
`
`892% mMEDGE
`
`03891
`
`
`as.5:
`
`
`
`m<mDan.6E9“.
`
`.mv
`
`5m952¢
`
`n:
`
`.8
`
`=mm
`
`...........................................................................
`
`..............................
`
`an?
`
`9
`
`
`
`

`

`U.S. Patent
`
`Aug. 31, 1999
`
`Sheet 9 of 10
`
`5,946,313
`
`ZZ
`
`
`
`9 |
`
`
`
`
`
`
`
`
`
`
`
`
`
`Z80982
`
`10
`
`

`

`U.S. Patent
`
`Aug. 31, 1999
`
`Sheet 10 0f 10
`
`5,946,313
`
`Z->
`
`9
`
`82
`
`
`
`
`
`0789
`
`#72
`
`ZZZZ
`
`11
`
`

`

`1
`MECHANISM FOR MULTIPLEXING ATM
`AAL5 VIRTUAL CIRCUITS OVER
`ETHERNET
`
`5,946,313
`
`2
`a portion of a number identifying the organization admin
`istering the remainder of the number, which is IEEE for
`Ethernet. The ID is a unique identifier that a manufacturing
`organization can provide to all equipment that it manufac
`tures. Further, this identifier is unique and Statically assigned
`and well known to the Station. Certain Ethernet addresses are
`reserved for broadcast and multicast to all end-stations on
`the Segment and for diagnostic purposes.
`An Ethernet connected end Station receives all data broad
`cast onto the media. By convention, the end Station discards
`all traffic not directed to itself, all, or a Subset of end Stations,
`as identified in the destination MAC address.
`All major emerging communication technologies rest on
`the layers of the OSI model. The OSI model defines a
`physical layer which specifies the Standards for the trans
`mission medium, a data link layer (layers 2 and 3) and a
`network layer (layers 4 to 7). Thus, in many cases, Ethernet
`operates on FDDI (fiber distributed data interface) physical
`layer, and the MAC layer, placed on top of FDDI, comprises
`the data layer. ATM operates on SONET, copper, twisted
`pairs, FDDI as physical layer, and the data layer is Subdi
`vided into an ATM layer and an ATM adaptation layer
`(AAL) providing the convergence function (called also
`convergence sublayer CS). Whatever the implementation of
`the AAL at the UNI, the ATM network is not concerned with
`the AAL operations, the ATM bearer service is masked from
`the convergence function.
`It has become evident that LAN shared bus architecture is
`insufficient to meet the demands of applications that require
`more bandwidth, and that LANs are beginning to become a
`bottleneck in computing environments. For this reason,
`more economic local interfaces Such as a Frame-Relay
`version (FUNI) and an Ethernet version, Cells-in-Frames
`(CIF), are used in the access network. In both cases, the
`Separation of data into cells is deferred until within the
`network, but the higher level information is carried to the
`end Station. In addition, according to the CIF version the
`AAL5, PDUs are pre-packaged at the end Station and this
`implies changes in HW and SW at each Ethernet connected
`end Station.
`Switched Ethernet technology, developed to provide more
`capacity to an end-user, does not relay on Shared medium, it
`rather provides point-to-point bandwidth between the user
`station and the Switch, so that instead of sharing a 10 Mbit/s
`medium, the user gets a dedicated 10 Mbits/S medium. AS
`Ethernet hubs and Switches are growing in use, they become
`an inexpensive means to provide more bandwidth to work
`stations. A Switched Ethernet network is more flexible, in
`that it may include Stations that are using a port at a given
`full rate, Stations that share a port, or Stations that have
`access to more than one port.
`However, Switched Ethernet provides only limited band
`width and Supports data traffic only. A more efficient Solution
`for bursty traffic is needed. There is also a need to simplify
`and Standardize the access link while also obtaining protec
`tion of the access traffic.
`Although ATM provides a very rich environment with
`numerous traffic classes and the ability to multiplex many
`data Streams with different handling requirements together,
`this functionality is mainly required in the network back
`bone. It is sought that ATM networks will be used by more
`general class end Stations for delivering multi-media Ser
`vices. However, in the short term, the extra bandwidth and
`cost of ATM interfaces is probably not justified for general
`class end Stations, Such as desktop computers. It is possible
`to built ATM Switches with lower speed ATM interfaces, but
`
`15
`
`FIELD OF THE INVENTION
`The invention is directed to a system and method for
`carrying ATM network information over a local area net
`work (LAN), and more particularly to a mechanism for
`multiplexing ATM AAL5 virtual circuits over Ethernet.
`BACKGROUND OF THE INVENTION
`The asynchronous transfer mode (ATM) forms the basis
`for Switching in broadband networks. ATM is a connection
`oriented data transport which is media independent. The key
`feature of ATM is the segmentation of data into fixed length
`units of data referred to as cells. Each cell is separately
`steered at each ATM Switch via an identifier of local sig
`nificance to the local transport leg provided in the header of
`each cell. The identifiers are reassigned during the transit of
`a cell from an input port to an output port on a Switch. The
`identifier carried between a Switch and an end System is 24
`bits in length. For an ATM user network interface (UNI), this
`is a concatenation of a 16 bit virtual circuit identifier (VCI)
`and an 8 bit virtual path identifier (VPI).
`This routing mechanism differS Significantly from other
`networks, in that there is only one identifier Specifying a
`local path VS. Source and destination information. This path
`information in itself is insufficient to uniquely identify the
`real Source and destination for the payload and therefore the
`connection is Set up via Signalling. AS Such, connection to a
`remote end-station is requested and upon connection Set-up,
`the network informs the end station what the local identifier
`of the connection is.
`Ethernet is a connectionless LAN technology designed for
`data applications in which all Stations on the network Share
`the communication medium. This medium, which could be
`twisted pairs, fiber, or coaxial cables, is shared in a peer to
`peer fashion. All devices on the Ethernet can be reached by
`a single transmission of data. Ethernet operates typically at
`10 Mbs and the data are sent in the form of Ethernet
`“frames'.
`There is no central arbitrator of bandwidth to administer
`media access on an Ethernet. Every time an Ethernet end
`Station sends message, it listens to the media to ensure that
`it is not in use by another Station. If this is true, the end
`Station commences Sending its own message. During the
`message Send phase, the end Station monitors the media to
`detect if another Station has also commenced Sending at the
`Same time. The minute delays imposed by the Speed of light
`permit a relatively large window wherein multiple Stations
`can believe that the media is idle, and therefore can com
`mence Sending an Ethernet frame. If the end Station detects
`a collision, i.e. what it hears does not match what it sends,
`it Switches to Sending a short “jabber Sequence to ensure
`that all colliding end Stations detect that contention has
`occurred. All end-stations detecting a collision will wait a
`random interval and will then retry Sending their frame, once
`again applying the Same rules to determine Success, and to
`free up the channel as quickly as possible when a collision
`occurs. Additional error detection is built into each frame to
`ensure that errored frames are not propagated.
`Ethernet end Stations are addressed globally and uniquely
`by a 48 bit media access control (MAC) address. The MAC
`address is comprised of a 24 bit Organization Unique
`65
`Identifier (OUI) and a 24bit end station identifier (ID). OUI
`is a globally administered numbering plan which comprises
`
`45
`
`50
`
`55
`
`60
`
`25
`
`35
`
`40
`
`12
`
`

`

`5,946,313
`
`15
`
`3
`this Solution presents a Serious deployment problem in that
`it requires replacement of the Substantial installed base of
`shared media LAN wiring and adapter cards.
`An ATM-Ethernet concentrator is disclosed in U.S. Pat.
`No. 5,457,681 (Gaddis et al., issued on Oct. 10, 1995 and
`assigned to Washington University), which provides an
`interface between an ATM network and a plurality of
`Ethernet Segments. Each Ethernet frame transmitted by any
`of the Ethernet Segments is fragmented into a Sequence of
`ATM cells, which are transmitted by an Ethernet controller
`asSociated with the respective Segment over the ATM net
`work and delivered to the interconnected Ethernet control
`lers. When the cells are received, the controller re-assembles
`them into frames and transmits the frames over the respec
`tive Ethernet segment to the end stations. While this patent
`partially addresses the problems of bandwidth and cost, it
`does not provide a method and system for transmitting ATM
`cells in Ethernet frames, for taking advantage of the ATM
`capabilities.
`There is a need to provide an improved network commu
`nication System with minimal displacement of existing net
`work components, capable of providing large bandwidth to
`the end Stations for data, Video and Voice traffic, and
`providing LAN access to Switched point-to-point WAN
`25
`linkS.
`International application No. PCT/CA95/00029 (WO
`95/20282) (by Burwell et al. published on Jul. 27, 1995 and
`assigned to Newbridge Networks Corporation) discloses a
`communication network comprising ATM Switches inter
`faced with LANs, the ATM cells being encapsulated in LAN
`frames and being delivered in encapsulated form over the
`Ethernet LAN direct to the end station. In another
`embodiment, the LAN interface adapter of the end station
`provide bridging, network layer functions and LAN emula
`tion functions to permit transparent communication between
`the end stations over the ATM network. The interface
`adapter, also defined as a “ridge (bridge/router) creates
`frames from ATM cells and vice-versa.
`However, the method disclosed in the above patent layers
`ATM carriage on top of the Ethernet layer. This is to say,
`ATM information only appears within the Ethernet payload,
`imposing an extra layer of indirection and frame processing
`on ATM handing at the LAN/WAN boundary.
`SUMMARY OF THE INVENTION
`It is an object of this invention to provide a mechanism for
`transmitting ATM cells in Ethernet frames for interworking
`of ATM backbone networks with the large base of legacy
`equipment, and for re-establishing access to an ATM net
`work for a LAN.
`It is another object of the invention to provide an address
`ing convention for carriage of ATM over Ethernet to a
`Specified end Station.
`This invention is based on the fact that although an
`Ethernet MAC address is normally globally unique for
`universal interoperability, it does not absolutely have to be
`for a dosed Ethernet to work. Uniqueness of the address
`within the Ethernet broadcast domain itself is necessary.
`According to this invention, an Ethernet end Station is
`allowed to assume multiple MAC level identifiers on a
`single Ethernet interface. These identifiers which are
`dynamically assigned, have only transitory Significance on
`the Ethernet.
`Accordingly, the invention provides a multiplexer
`(E-Mux) for encapsulating ATM cells into a LAN frame,
`comprising a Segmentation and reassembly unit for receiv
`
`35
`
`4
`ing a plurality of incoming ATM cells with a LAN destina
`tion address and generating an ATM adaptation layer 5
`(AAL5) protocol data unit (PDU); a PDU manager for
`receiving the AAL5 PDU and extracting an AAL5 payload;
`a header processor for extracting a traffic type indicator from
`the header of the PDU; an address processor for extracting
`the LAN destination address from the header of an incoming
`ATM cell; and a frame manager for receiving the traffic type
`indicator, the AAL5 payload and the LAN destination
`address and generating an incoming LAN frame.
`The invention also provides a multiplexer (E-Mux) for
`segmenting a LAN frame into a plurality of ATM cells,
`comprising: a frame manager for receiving an outgoing
`LAN frame with an ATM destination address and
`de-assembling it into a traffic type indicator, an AAL5
`payload and an ATM destination address, a header processor
`for receiving the traffic type indicator from the frame man
`ager; a PDU manager for receiving the AAL5 payload and
`the traffic type indicator and generating an ATM adaptation
`layer 5 (AAL5) protocol data unit (PDU); an address pro
`cessor for receiving the ATM destination address from the
`frame manager; and a Segmentation and reassembly unit for
`receiving the PDU and the ATM destination address gener
`ating a plurality of ATM cells with the ATM destination
`address.
`According to another aspect of the invention, there is
`provided a telecommunication network comprising a LAN
`with a plurality of end-Stations connected over a transmis
`Sion medium and an ATM network, comprising: a multi
`plexer (E-Mux) for encapsulating a plurality of ATM cells
`received from the ATM network into an incoming LAN
`frame and for Segmenting a LAN frame received from the
`LAN network into a plurality of outgoing ATM cells; an
`E-Mux-to-LAN interface for adapting the transmission for
`mat of the LAN frame for transmission over the connection
`medium of an LAN network; an ATM-to-E-Mux interface
`for adapting the transmission format of the ATM cells
`received from an ATM network for processing by the
`E-Mux; an address controller at each end station for for
`warding the incoming LAN frame to the end Station when a
`destination address comprised in the destination MAC field
`of the LAN frame is recognized by the address controller,
`and for inserting an ATM destination address into the Source
`MAC field of the outgoing frame.
`A method for transmitting information from an ATM
`network to an Ethernet network using an E-Mux is also
`disclosed, the method performing the Steps of establishing
`connection between an ATM Switch of the ATM network and
`an end station of the Ethernet network based on a VPI/VCI
`destination address in the header of an incoming ATM cell;
`receiving a plurality of incoming ATM cells with the desti
`nation address, and generating an ATM adaptation layer 5
`(AAL5) protocol data unit (PDU) with a segmentation and
`reassembly unit; extracting an AAL5 payload from the PDU
`with a PDU manager; extracting a traffic type indicator from
`the header of the PDU with a header processor; generating
`with a frame manager an incoming Ethernet frame using the
`traffic type indicator, the AAL5 payload and the destination
`address, and transmitting the Ethernet frame over the Eth
`ernet network to the end Station according to the destination
`address.
`According to Still another aspect of the invention, there is
`provided a method for transmitting information from an
`Ethernet network to an ATM network using an E-Mux
`performing the Steps of establishing connection between an
`end station of the Ethernet network and an ATM Switch of
`the ATM network based on a VPI/VCI destination address
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`13
`
`

`

`S
`provided by the end Station; generating an outgoing Ethernet
`frame at the end Station and transmitting Same to the E-Mux;
`extracting from the Ethernet outgoing frame a traffic type
`indicator, a frame payload and a Source address, with a frame
`manager; generating an ATM adaptation layer 5 (AAL5)
`protocol data unit (PDU) with a segmentation and reassem
`bly unit from the frame payload and a traffic type indicator
`extracted from the type field of the outgoing frame with a
`header processor, Segmenting the PDU into a plurality of
`outgoing ATM cells and inserting a VPI/VCI destination
`address in the header of the cells from an address processor.
`Advantageously, the System and method of the invention
`provides a highly efficient carriage of ATM information to an
`end Station using actual MAC address Space in the Ethernet
`frame.
`Because the frames are not pre-packaged into AAL5
`PDUs at the end station, but at the E-Mux, the system of the
`invention is efficient, as custom hardware to perform this
`function does not need to be deployed at each Ethernet
`connected end Station, and the end Station Software is not
`burdened with this task. Custom hardware to perform SAR
`(Segmentation and reassembly) function is built into the
`E-MuX.
`Still another advantage of the invention is that legacy
`25
`Ethernet can coexist with ATM LAN to UNI traffic. ATM
`intelligence can be distributed in Such a LAN Segment. For
`meshed PVC/UBR type connections, standard packet for
`mats and driver interfaces may be used between layer 3 and
`the Ethernet interface, making the invention applicable to
`various LAN technologies. No additional information needs
`to be propagated outside the System of the invention.
`Therefore, this permits the System to be tailored as an
`application specific ATM to Ethernet interface.
`Still another advantage of the invention is that the Eth
`ernet frame is efficiently used since the ATM routing infor
`mation is embedded in the Ethernet MAC addressing field.
`No additional ATM header within the Ethernet payload is
`necessary, Since the entire ATM SemanticS is not carried to
`the end Station. This does not preclude, however, additional
`Semantics being packaged in a separate header in the Eth
`ernet frame.
`ATM virtual circuit (VC) address mapping is carried
`forward into the Ethernet domain. Intermediate Steps are not
`required to map Ethernet MAC to VC, as VC information
`flows end-to-end. For uncommitted bit rate (UBR) data
`services, only the ATM path identifier need to be carried
`forward across an Ethernet LAN.
`As a result, Ethernet traffic can coexist-exist with ATM
`LAN to UNI traffic. AS indicated above, each Ethernet
`50
`connected Station may assume multiple IDS on the LAN at
`the MAC level. Thus, for traditional, non-ATM traffic, it
`retains the manufacturers MAC address built into the Eth
`ernet interface at the factory. For ATM traffic, it assumes one
`or more IDs depending on the ATM virtual circuits which
`ultimately terminate at that particular end Station interface.
`
`1O
`
`15
`
`35
`
`40
`
`45
`
`55
`
`5,946,313
`
`6
`FIG. 1C shows an Ethernet frame (prior art);
`FIG. 2 shows the block diagram of the E-Mux;
`FIG.3A shows the processing of the ATM cells to produce
`an incoming Ethernet frame;
`FIG. 3B Shows the processing of an outgoing Ethernet
`frame to produce ATM cells;
`FIG. 4A is a the flow-chart illustrating the assembly of the
`ATM cells to produce an Ethernet frame;
`FIG. 4B is a the flow-chart illustrating the segmentation
`of the Ethernet frame to produce ATM cells;
`FIG. 5 shows the block diagram of a variant of the E-Mux;
`FIG. 6A shows the processing of the ATM cells to produce
`an incoming Ethernet frame according to the variant of FIG.
`5; and
`FIG. 6B Shows the processing of an outgoing Ethernet
`frame to produce ATM cells according to the variant of FIG.
`5.
`
`DESCRIPTION OF THE PREFERRED
`EMBODIMENT
`FIGS. 1A, 1B, and 1C are provided for defining and
`illustrating Some of the terms necessary for describing the
`present invention and its mode of operation.
`FIG. 1A Shows the convergence function, and the Seg
`mentation and reassembly (SAR) function of ATM. The
`convergence function is responsible for accepting the user
`traffic which could range from one to maximum 65,000
`Bytes, and placing a header 10 and a trailer 12 around it to
`obtain a protocol data unit (PDU) 2. For this invention, the
`payload field of the PDU is limited to 1,500 bytes, which is
`the size of the Ethernet payload field. To push PDU beyond
`Ethernet limit of 1,500 bytes will require an additional
`header in the Ethernet frame. The length of the header and
`trailer is between 6 and 40 bytes. Once the header and the
`trailer have been added to the user payload, the traffic is
`segmented into 44-48 bytes data units 14. Next, the adap
`tation layer adds a header 16, and possibly a trailer 18 to the
`data unit 14, depending on the type of payload being
`Supported. In any event, the final data unit from this opera
`tion is always a 48 octet block 20. Finally, the last operation
`is performed by the data link layer which adds a five-octet
`header 22 to the 48-octet payload 20 resulting in a 53 bytes
`cell 24. Each cell is transported over the physical layer
`between two ATM Switches designated by the address
`information in header 22.
`FIG. 1B shows a basic linear point-to-point ATM network
`configuration where the connections are identified through
`virtual channel identifiers (VCI) and virtual path identifiers
`(VPI) in the ATM cell header. Switching in the ATM
`network is illustrated at 5, 7, and 9. A virtual channel
`connection (VCC) 11 has end-to-end significance between
`end users A and B. A virtual path connection (VPC) has
`significance between adjacent ATM devices, 5, 7, and 9, and
`Switching is performed very quickly through the use of a
`routing table.
`FIG. 1C illustrates an Ethernet frame. Ethernet end sta
`tions are addressed globally and uniquely by the MAC
`address. Field 28 comprises the destination MAC and field
`30 comprises the source MAC. The MAC address is has a 24
`bit organizationally unique identifier (OUI) 38, 42 and a 24
`bit end station identifier 40, 44.
`A type field 32 is provided for specifying the traffic type.
`The payload field 34 may comprise up to 1500 bytes. The
`frame begins with a training Sequence 26 for allowing
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`The foregoing and other objects, features and advantages
`of the invention will be apparent from the following more
`particular description of the preferred embodiments, as
`illustrated in the appended drawings, where:
`FIG. 1A Shows the convergence and Segmentation and
`reassembly functions of ATM (prior art);
`FIG. 1B shows an ATM network for defining the terms
`Virtual channel and virtual path (prior art);
`
`60
`
`65
`
`14
`
`

`

`5,946,313
`
`1O
`
`15
`
`25
`
`35
`
`40
`
`7
`receiver Synchronization and ends with a frame check
`Sequence 36, for determining the integrity of the data in the
`frame.
`FIG. 2 shows the block diagram of the System according
`to the invention. An E-Mux 21 exchanges ATM cells with an
`ATMUNI 27, which is connected in turn to an ATM network
`23. E-Mux 21 is also connected to an Ethernet LAN 25 for
`eXchanging frames. By convention, the traffic travelling
`from ATM network 23 to LAN 25 is defined by the term
`“incoming”, and the traffic travelling from LAN 25 to ATM
`network 23 is defined by the term “outgoing”.
`According to the invention, a unique MAC OUI is estab
`lished for extending ATM path addressing into the Ethernet
`MAC address domain. This unique MAC OUI identifies the
`traffic as ATM UNI. The ATM OUI is inserted in both the
`destination and the Source MAC fields 38 and 42. The OUI
`field informs LAN 25 that the traffic is coming from a source
`not registered to it, So as to treat it accordingly. AS indicated
`earlier, this permits the ATM traffic to coexist with tradi
`tionally addressed Ethernet traffic.
`E-Mux 21 associated to LAN25 also requires a unique ID
`in the destination MAC, Such that incoming traffic can be
`uniquely addressed to it. The ID could be for example
`VPI=0, VCI =0, which is never used in the ATM network 23.
`This unique address, ATM OUI, VPI=0,VCI=0, is mapped
`into the source MAC field 30 of an incoming frame and in
`the destination MAC of an outgoing frame. This is described
`next in connection with FIG. 3A, which shows the process
`ing of the ATM cells to produce an incoming Ethernet frame
`and FIG. 3B, which shows the processing of an outgoing
`Ethernet frame to produce ATM cells.
`Each end Station has an VPI/VCI address for the ATM
`traffic used to establish connection between a Switch in the
`ATM network and an end Station through Signalling in a
`known manner. For the incoming traffic, a flow of cells 24,
`addressed to an end station 39, 39', 39" in Ethernet network
`25 is received at UNI 27 from network 23. The cells are
`assembled into a PDU 2 by SAR unit 29 of E-Mux 21.
`PDU 2 comprises a AAL5 payload field 14 for receiving
`the payload from the incoming cells 24, the size of the
`payload field being limited to the maximum size of field 34
`(up to 1500 bytes) of an Ethernet frame 3. Although the
`normal ATM AAL5 protocol data unit (PDU) is quite large,
`as discussed in connection to FIG. 1, it can be constrained
`to fit within the 1500 bytes of an Ethernet frame.
`Since the size of the PDU is restricted to the payload field
`length of a frame, the cell Segmentation function is per
`formed by SAR 29 at boundary to the ATM network, such
`that the need for additional PDU information to be carried in
`the payload portion of the Ethernet frame is obviated. This
`differs from the cell-in-frame (CIF) approach in which the
`SAR function is performed at the end station, and then the
`cells are reassembled into Ethernet frames for transmission.
`A PDU manager 31 strips the LLC/SNAP (Logical link
`control, sub-network attachment point) header of the PDU 2
`and provides it to a header processor 37, which determines
`the type of the payload. The payload and the payload type
`are forwarded to a frame manager 33.
`Frame manager 33 generates the Ethernet frame 3 by
`mapping the payload into field 34 and the payload type into
`field 32. Frame manager 33 also receives the address of the
`destination end Station from an address processor 35, and
`maps this information into field 40 of destination MAC. This
`address is a concatenation of the VPI/VCI address extracted
`from the cell header 22. As well, address processor 35 maps
`the address of the E-Mux in the Source MAC field 30,
`
`45
`
`50
`
`55
`
`60
`
`65
`
`8
`namely ATM OUI and VPI=0, VCI=0 in fields 40 and 42,
`respectively. AS Such, for the incoming traffic, the Source
`MAC field of frame 3 comprises the address of the E-Mux,
`and the destination MAC comprises the address of the end
`station in the Ethernet.
`Frame manager 33 sends frame 3 assembled as indicated
`above, over the Ethernet network 25. An interface 47 is
`provided for adapting the format of the frame to the con
`nection medium of Ethernet 25.
`In general, each “ATM aware” end station 39, 39', 39" is
`provisioned with an address controller, as shown at 41, 41',
`41", which allows the end station to signal the network to
`request/accept connections. AS well, address controller 41,
`41', 41" provides the range of VPI/VCI values that could be
`assigned for Such connections to the associated end Station
`39, 39', 39 to a defined subset of the whole VPI/VCIs
`allocated for UNI 27. An address controller 41, 41', 41"
`recognizes a destination ID as being its own, using a look-up
`table, and directs frame 3 to the associa