Date post: | 28-Apr-2019 |
Category: |
Documents |
Upload: | hoangtuong |
View: | 217 times |
Download: | 0 times |
HUAWEI TECHNOLOGIES Co., Ltd.
PBL Solution for HSE
Trey MalpassZeng LiYe Min
IEEE 802.3ba Task Force, 23-25 Jan 2008
HUAWEI TECHNOLOGIES Co., Ltd.
Page 2
IEEE 802.3ba Task Force, 23-25 Jan 2008
PBL Model
Block Generator
Block Distributor 64B/66B
encoder64B/66Bencoder
64B/66Bencoder
66bitLane1
Lane2
Lane10
CTBI
VCSEL Array
CGMII
BlockProcess
Block Alignment
64B/66Bdecoder64B/66Bdecoder
64B/66Bdecoder
66bitLane1
Lane2
Lane10
Alignment BlockGenerator
Alignment BlockDetector
PIN
PIN
PIN
PCS PMA/PMDMAC/RS
Buffer
Bit Matrix
Bit Matrix
N x(64+4)bit
N x(64+4)bit
HUAWEI TECHNOLOGIES Co., Ltd.
Page 3
IEEE 802.3ba Task Force, 23-25 Jan 2008
PBL Model
• Receives data from MAC layer, generates 64-bit blocks• Transmits blocks through the CGMII bus• Blocks distributed to N lanes, alignment words will be
inserted in each lane following the distributor.• Encode the blocks in each lane (64B/66B encoding &
scrambler)• Bit matrix will adapt N-lane structure to standard CTBI for
all PMDs• Preserves block format in the physical Lanes (Block
interleave in physical lanes)
HUAWEI TECHNOLOGIES Co., Ltd.
Page 4
IEEE 802.3ba Task Force, 23-25 Jan 2008
Block Generator and CGMII Interface
• MAC data frame will generate the 64-bit blocks using the existing 64B/66B rule.
• Based on the type of the Block, RS generates the 4-bit control code signal.
• (64+4)bit is the unit of the CGMII for each lane
• PBL distributes the blocks to each lane
Block GeneratorControl Signal
Generator
RS
MACMAC Data Frame
TXC<3:0> TXD<63:0>
Lane 0Lane 1Lane n
Distributed to Lane
Idle Insert
CGMII
HUAWEI TECHNOLOGIES Co., Ltd.
Page 5
IEEE 802.3ba Task Force, 23-25 Jan 2008
Block Distribution and Alignment Insertion
•Alignment blocks will be added into the stream by increasing lane rate
Block Generator
Block Distributor 64B/66B
encoder64B/66Bencoder
64B/66Bencoder
66bitLane1
Lane2
Lane10
CTBI
Alignment BlockGenerator
PCSMAC/RS
scramble
scramble
scramble
•••Packet IPG Packet IPG
64B
64B
64B
A
A
A
64B 64B 64B 64B
64B 64B 64B 64B
64B 64B 64B 64B
•••
64B 64B64B ••• 64B64B
66B
66B
66B
A
A
A
66B 66B 66B
66B 66B 66B
66B 66B 66B
•••
Alignment block after 64/66 encoding
Alignment block before 64/66 encodingA
A
A
A
A
64B
64B
64B
A
A
A
HUAWEI TECHNOLOGIES Co., Ltd.
Page 6
IEEE 802.3ba Task Force, 23-25 Jan 2008
Insert alignment block by increasing lane rate
• Alignment block is defined using a 64B/66B block format.
• Increase the lane rate to insert the alignment blocks.– If we send an alignment block once every 1375 blocks, the lane rate will
be 10.32GHz (0.073% increase)– If we send an alignment block once every 16k blocks, the lane rate will be
10.313GHz (0.00625% increase)– Alignment block rate will affect
• the maximum skew that can be compensated• adaptation to OTN • receive RAM size for de-skew• clock rate generator complexity in the encoding modules
HUAWEI TECHNOLOGIES Co., Ltd.
Page 7
IEEE 802.3ba Task Force, 23-25 Jan 2008
What should the alignment block look like
•Alignment block:64-bit data + 4-bit ctrl
•Alignment block after 64/66 coding:
–SEQ is alignment block sequence number
–ID indicates the lane number
5a SEQ ID x x x x x1101
5a SEQ ID x x x x x10
TXD/RXD TXC/RXC DescriptionD D D D D D D D 0000 DataS D D D D D D D 1001 StartC C C C S D D D 1010 StartT C C C C C C C 1000 TerminateD T C C C C C C 0111 TerminateD D T C C C C C 0110 TerminateD D D T C C C C 0101 TerminateD D D D T C C C 0100 TerminateD D D D D T C C 0011 TerminateD D D D D D T C 0010 TerminateD D D D D D D T 0001 TerminateC C C C C C C C 1111 ControlA A A A A A A A 1101 AlignmentO O O O O O O O New OAMN N N N N N N N New NullX X X X X X X X New Reserve
HUAWEI TECHNOLOGIES Co., Ltd.
Page 8
IEEE 802.3ba Task Force, 23-25 Jan 2008
Insert alignment block by increasing lane rate
•The received data is unaligned because of the skew introduced across physical lanes•Decode and detect the alignment block•Realign using alignment block with the same SEQ•Reassemble data steam in order by lane ID.
64B
64B
64B
A
A
A
64B 64B 64B 64B
64B 64B 64B 64B
64B 64B 64B 64B
•••
66B
66B
66B
A
A
A
66B 66B 66B 66B
66B 66B 66B 66B
66B 66B 66B 66B
•••
64B
64B
64B
A
A
A
64B 64B 64B 64B
64B 64B 64B 64B
64B 64B 64B 64B
•••
64B64B64B64B64B •••
Skewed data
A
A
A
•••
64/66 decode
Re-alignment Re-combine
HUAWEI TECHNOLOGIES Co., Ltd.
Page 9
IEEE 802.3ba Task Force, 23-25 Jan 2008
PHY OAM Process
• PHY OAM process can be used to monitor lane failures.– LLF/RLF
– Others, i.e. Auto negotiate
• PHY OAM code is defined as a new block type.• Insert PHY OAM by stealing from IPG
HUAWEI TECHNOLOGIES Co., Ltd.
Page 10
IEEE 802.3ba Task Force, 23-25 Jan 2008
What should the OAM block look like•OAM block: 64-bit data + 4-bit ctrl TXD/RXD TXC/RXC Description
D D D D D D D D 0000 DataS D D D D D D D 1001 StartC C C C S D D D 1010 StartT C C C C C C C 1000 TerminateD T C C C C C C 0111 TerminateD D T C C C C C 0110 TerminateD D D T C C C C 0101 TerminateD D D D T C C C 0100 TerminateD D D D D T C C 0011 TerminateD D D D D D T C 0010 TerminateD D D D D D D T 0001 TerminateC C C C C C C C 1111 ControlA A A A A A A A 1101 AlignmentO O O O O O O O 1011 OAMN N N N N N N N New NullX X X X X X X X New Reserve
1011
Block type:8bits
OAM type: 8bits•x01H:LLF•x02H:RLF•x03H:status advertisement•x04H:BER Monitor•Others: reserved
Broadcast OAM / lane OAM : 2bits
lane ID : 6bits
OAM value : 24bits
reserved
0xa5
HUAWEI TECHNOLOGIES Co., Ltd.
Page 11
IEEE 802.3ba Task Force, 23-25 Jan 2008
Insert OAM block by stealing from IPG
Block Generator
Block Distributor 64B/66B
encoder64B/66Bencoder
64B/66Bencoder
66bitLane1
Lane2
Lane10
CTBI
OAM BlockGenerator
PCSMAC/RS
scramble
scramble
scrambleIPG DC
•••
• IPG adjustment gains gains enough bandwidth for inserted OAM• Introduce Deficit Counter mechanism to collect extra IPG• OAM blocks cover the extra IPG (generator)
Packet IPG Packet IPG Packet IPG Packet IPG Packet IPG Packet PacketIPG IPG Packet
64B 64B 64B64B64B64B64B64B ••• ••• 64B64B64B •••64B •••
IPG Packet IPG Packet
HUAWEI TECHNOLOGIES Co., Ltd.
Page 12
IEEE 802.3ba Task Force, 23-25 Jan 2008
Insert OAM block by stealing from IPG
64B
64B
64B
O
O
O
64B 64B 64B
64B 64B 64B
64B 64B 64B•••
66B
66B
66B
O
O
O
66B 66B 66B
66B 66B 66B
66B 66B 66B•••
OO OAM block after 64/66 encoding
OAM block before 64/66 encoding
64B
64B
64B
64B
64B
64B
64B 64B 64B
64B 64B 64B
64B 64B 64B•••
64B IPG Block
64B 64B64B64B64B ••• ••• 64B64B64B64B •••
• Block generator adjusts IPG to get IPG blocks in each lane• Distributor sends IPG blocks to each lane• PHY OAM replaces the IPG block
HUAWEI TECHNOLOGIES Co., Ltd.
Page 13
IEEE 802.3ba Task Force, 23-25 Jan 2008
Fault Indication by PHY OAM Block
Block Generator
Block Distributor 64B/66B
encoder64B/66Bencoder
64B/66Bencoder
66bitLane1
Lane2
Lane10
OAM BlockGenerator
PCSMAC/RS
scramble
scramble
scrambleIPG DC
Block Reassemble
Block Distributor 64B/66B
encoder64B/66Bencoder
64B/66Bencoder
66bitLane1
Lane2
Lane10
OAM BlockDetector
scramble
scramble
scramble
RLF
Receiver gets lane failure information, returns a Remote Link Fail status to Transmitter for OAM block generation
HUAWEI TECHNOLOGIES Co., Ltd.
Page 14
IEEE 802.3ba Task Force, 23-25 Jan 2008
Block Encoding
• Receive distributed blocks in each lane• 64/66 encoding format is compatible to 10GE• Independently encode and scramble in each lane• The scrambler polynomial
– G(X)=1+X39+X58
HUAWEI TECHNOLOGIES Co., Ltd.
Page 15
IEEE 802.3ba Task Force, 23-25 Jan 2008
64B/66B Block FormatsOriginal Data (64bit+4bit) Output Data (66Bit)
Description TXC/RXC TXD/RXD SYNC Block Payload
Data 0000 D D D D D D D D 01 D D D D D D D DControl Block Formats Block
Type
Start 1001 S D D D D D D D 10 0x78 D D D D D D DStart 1010 C C C C S D D D 10 0x33 C’ C’ C’ C’ D D D
Terminate 1000 T C C C C C C C 10 0x87 C’ C’ C’ C’ C’ C’ C’Terminate 0111 D T C C C C C C 10 0x99 D C’ C’ C’ C’ C’ C’Terminate 0110 D D T C C C C C 10 0xaa D D C’ C’ C’ C’ C’Terminate 0101 D D D T C C C C 10 0xb4 D D D C’ C’ C’ C’Terminate 0100 D D D D T C C C 10 0xcc D D D D C’ C’ C’Terminate 0011 D D D D D T C C 10 0xd2 D D D D D C’ C’Terminate 0010 D D D D D D T C 10 0xe1 D D D D D D C’Terminate 0001 D D D D D D D T 10 0xff D D D D D D D
Control 1111 C C C C C C C C 10 0x1e C’ C’ C’ C’ C’ C’ C’ C’Alignment 1101 A A A A A A A A 10 0x5a SEQ ID x x x x x
OAM 1011 O O O O O O O O 10 0xa5 OAMTYPE OAM Value
Null New N N N N N N N N 10 0x00 0x00Reserve New X X X X X X X X 10
HUAWEI TECHNOLOGIES Co., Ltd.
Page 16
IEEE 802.3ba Task Force, 23-25 Jan 2008
Bit Matrix
• Adaptation between N PBL lanes to CTBI• Adapts to any N-lane physical transport by bit transpose• Bit Matrix is (n x m) memory
– n is number of distributed lanes (physical lanes) – m is number of electrical lanes in CTBI signal (10)
HUAWEI TECHNOLOGIES Co., Ltd.
Page 17
IEEE 802.3ba Task Force, 23-25 Jan 2008
Bit Transpose (10:4)
64B/66Bencoder
64B/66Bencoder
64B/66Bencoder
4×25G
66bit
Lane1
Lane2
Lane4
CTBI
64B/66Bencoder
LD EML
LD EML
LD EML
LD EML
Lane3 MUXOptical Module
Bit skew
Bit Matrix block
bit
HUAWEI TECHNOLOGIES Co., Ltd.
Page 18
IEEE 802.3ba Task Force, 23-25 Jan 2008
Bit Matrix
• The block from the independent lane is written into the row of matrix bit by bit
• Bits will be read from column of the matrix• A de-skew lane will help re-align the bits across the CTBI
– The mechanism is the same as the SFI-5 definition.
• A bit matrix is used in the 4x25G module.
HUAWEI TECHNOLOGIES Co., Ltd.
Page 19
IEEE 802.3ba Task Force, 23-25 Jan 2008
Bit Pass Through
64B/66Bencoder
64B/66Bencoder
64B/66Bencoder
10×10G
66bit Lane1
Lane2
Lane10
CTBIBit Matrix
Special mode for 10x10G module, the pass through mode can keep the blocks running in the lanes
HUAWEI TECHNOLOGIES Co., Ltd.
Page 20
IEEE 802.3ba Task Force, 23-25 Jan 2008
What do the PMD Modules look like?
Bit Matrix
10X10Gmodule
PL1
PL4
PL2PL3
PL10
PL9
CTBI
XLFBI
PL1
PL4
PL2PL3
4X10Gmodule
When the number of nTBI lanes = the number of module physical lanes, then the module is very simple and 64/66 blocks are preserved at the final output.
•Nx10G PMDs
HUAWEI TECHNOLOGIES Co., Ltd.
Page 21
IEEE 802.3ba Task Force, 23-25 Jan 2008
What do the PMD Modules look like?
For the 4x25G PMD module, the bit matrix is required to go from 10-lane CTBI back to 4- lane physical; but this module is better able to tolerate the extra cost.
•4x25G PMD
64B/66Bencoder
64B/66Bencoder
64B/66Bencoder
4×25G
66bit
Lane1
Lane2
Lane4
CTBI
64B/66Bencoder
LD EML
LD EML
LD EML
LD EML
Lane3 MUXOptical Module
Bit skew
Bit Matrix block
bit
HUAWEI TECHNOLOGIES Co., Ltd.
Page 22
IEEE 802.3ba Task Force, 23-25 Jan 2008
Applications with PBL Model
• Backplane Transmission• Mapping to OTN
HUAWEI TECHNOLOGIES Co., Ltd.
Page 23
IEEE 802.3ba Task Force, 23-25 Jan 2008
Backplane Application•Backplane transmission
BlockBit
XLFBI
4X10G
PL1
PL4
PL2
PL3
Distributor
FEC
FECFEC
FEC
PMA/PMD
4x10GBASE-KR
FEC function is implemented in each lane; re-use existing KRCurrent FEC requires preservation of 64/66 block boundaries
HUAWEI TECHNOLOGIES Co., Ltd.
Page 24
IEEE 802.3ba Task Force, 23-25 Jan 2008
OTN application•PBL in OTN application
Bit Matrix
10X10Gmodule
PL1
PL4
PL2PL3
PL10
PL9
Transcode
Transcode
Transcode
OD
Un-Xv
OTN Line Card
CTBI SFI
XLFBI
PL1
PL4
PL2PL3
4X10Gmodule
Transcode
Transcode
Transcode
OD
Un-Xv
SFI
Transcode
10X10Gmodule
4X10Gmodule
CTBI
XLFBI
HUAWEI TECHNOLOGIES Co., Ltd.
Page 25
IEEE 802.3ba Task Force, 23-25 Jan 2008
OTN application•PBL in OTN application
CTBIPL1
PL4
PL2PL3
4X25G Module with CTBI Skew Line
Transcode
Transcode
Transcode
OD
Un-Xv
SFI
Transcode
OTN Line Card
CEI-25G
PL1
PL4
PL2PL3
4X25G Module with CEI-25G(In the future)
Transcode
Transcode
Transcode
OD
Un-Xv
SFI
Transcode
HUAWEI TECHNOLOGIES Co., Ltd.
Page 26
IEEE 802.3ba Task Force, 23-25 Jan 2008
Conclusion
• Blocks running in Physical lanes (rather than bit interleave)– Benefit to OTN mapping and Backplane applications by using existing
technology.– Enables fault detection and maintenance per physical lane– Physical Bundling Layer can be easily extended to higher-rate Ethernet in the
future (1Tbps…)– Block integrity in each physical lane is an important feature
• Alignment blocks will be needed in multi-lane formats– Define the alignment block by extending 10GE block formats– Insert alignment blocks into the data steam
• PHY OAM provides the monitor of physical lane– Lane failure will be indicated (LLF/RLF)– More functions can be considered using PHY OAM (Auto Negotiate, BER monitor)