CN1235357C - Full optical packet switching node structure for supporting burst or non-burst businesses - Google Patents
Full optical packet switching node structure for supporting burst or non-burst businesses Download PDFInfo
- Publication number
- CN1235357C CN1235357C CNB031295371A CN03129537A CN1235357C CN 1235357 C CN1235357 C CN 1235357C CN B031295371 A CNB031295371 A CN B031295371A CN 03129537 A CN03129537 A CN 03129537A CN 1235357 C CN1235357 C CN 1235357C
- Authority
- CN
- China
- Prior art keywords
- delay line
- input
- wave
- burst
- packet
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Abstract
The present invention relates to a full optical packet switching node structure for supporting the burst or non-burst business. A symmetrical structure is composed of a plurality of (n) 1*W light wave decomposing multiplexers, a plurality of (n) W*1 optical wave-division multiplexers, a nonblocking full optical switching matrix module (WN+R+M) *(WN+R+M), a plurality of (R) full wavelength shifters, a plurality of (M) optic fiber delay lines, wherein the N is provided with the number of input/output optical fiber ports. The W is provided with the number of wave-length channels which can be transmitted at each port, and a simple and degenerate type structure is composed of the (M) delay lines. The shortest or the longest delay line respectively provides the buffer-storing time T or MT to the packet. After the packet loaded by each wave length on the optical fiber is demultiplexed by the light wave decomposing multiplexer, the space division switching is carried out in the switching matrix. The input packet is switched to the destination port or is fed back to the shared wavelength shifter or the optic fiber delay line to solve the impact of the destination port of the packet. The present invention has the capability to greatly solve the burst or non-burst business and reduces the cost and volume of a switching system.
Description
Technical field:
The present invention relates to a kind of full optical packet switching node structure, solve the destination interface collision problem of light grouping.This switching fabric can be applicable to non-burst service (as the business of legacy circuit-switched character) and burst service (as IP operation) environment, belongs to the optical communication technique field.
Background technology:
At present, make communication network just gradually by based on circuit switching based on developing rapidly of the internet of IP, the mode of optimizing the voice-over business is to based on packet switching, and the mode of optimizing bearing data service develops.The application of video request program, conferencing over ip and other various novel multimedia services is had higher requirement to the transfer capability of communication infrastructure network.Dense wave division multipurpose that development in recent years is got up (DWDM) and erbium-doped fiber amplifier (EDFA) technology are very effective to the communication dilatation.But indispensable switching equipment is still electronic type (node still needs O/E, the E/O conversion and the signal of telecommunication to handle) in the communication network, thereby has seriously limited the further performance of optical fiber communication advantage.A kind of solution is directly to carry out light signal to handle, and promptly builds All-Optical Communication Network.
It can be line level, packet level or bit-level that light signal is handled.Wavelength division multiplexing (WDM) Optical Transmission Network OTN belongs to the light signal of line level to be handled, and is that the channel of coarseness is cut apart.Optical Time Division Multiplexing (OTDM) is that the light signal of bit-level is handled, because the requirement of the operating rate of sun adjuster spare is very high, also has suitable distance from practicability.Light packet switching (OPS) belongs to the light signal of packet level to be handled, and compares with the OTDM technology, and the requirement of sun adjuster spare operating rate reduces greatly, compares with WDM technology, can be with the quick allocated channel of thinner granularity, more flexibly, effectively utilize bandwidth.Therefore, the full light packet switching network has a extensive future.
Up to now, people have proposed various schemes, are used for solving the packet collisions problem of full optical packet switch network, may be summarized to be the following aspects:
Manage conflict from time-domain, (Fiber Delay Lines, FDLs) buffer memory competition packet is till the wavelength channel free time of institute's contention as adopting fibre delay line.According to the connected mode of FDL, can be divided into feed forward type (Feed Forward) and two kinds of connected modes of reaction type (Feedback) again.(as the M-Quadro structure, I.Chlamtac, A.Fumagalli, Chang-Jin Suh.Multibuffer delay line architecturesfor efficient contention resolution in optical switching nodes.IEEE Transactions onCommunications, 2000,48 (12): 2089-2098, and by structures that the people proposed such as K.Karol, K.Karol.Shared memory optical packet (ATM) switch.SPIE, Multigigabit FibreCommunications Systems, vol.41,1993:212-222.)
Solve competition from wavelength domain, promptly when the packet that transmits on a certain wavelength arrives switching system, if the wavelength of switching system output port correspondence is occupied, then by wavelength conversion with the carrying wavelength conversion of this packet to another idle wavelength, make packet on this idler long-channel, be sent to next node.
Solve competition from spatial domain, i.e. deflection routing mode, it utilizes the idle route in the network, makes packet walk around the circuitous destination node of going to of competition node.
The deflection routing mode is that other node and transmission link in the network are solved conflict as buffer, and its caused reception out of order meeting of dividing into groups is very serious, has also brought long time-delay in addition.And when heavier or network connection degree was low when offered load, the deflection routing mode was no longer valid to managing conflict.Therefore, the deflection routing mode is not fundamentally resolved the collision problem of switching node.Solve when competition from wavelength domain or from time-domain separately, also have following problem.
When the switching node in the traditional design utilized FDL to manage conflict, maximum problem was: (1) volume.For example: provide the buffer memory of 250 groupings at the 10Gb/s linear speed, for ATM cell, buffer memory need hold 53 * 8 * 250=106000 bit, is equivalent to the time-delay of 10.6 μ s, promptly needs to be similar to the FDL of 3km length; Switching equipment and connection device brought crosstalks and insert loss because of extra introducing when (2) realizing large capacity cache.
Therefore, people such as Danielsen propose to utilize the wavelength shifter (S.L.Danielsen that manages conflict purely at switching node, C.Joergensen, B.Mikkelsen, and K.E.Stubkjaer.Optical packetswitched network layer without optical buffers.IEEE Photonics Technology Letters, 1998,10 (6): 896-898.), from its research as can be known: when the number of wavelengths on the every optical fiber is no less than 8, being 0.8Erlang for every fine load, grouping arrives, and discarded ratio of packet can be less than 1,000,000/.But the problem of this structure maximum is that used wavelength shifter quantity is too many, reaches NW (N is the input-output optical fiber port, and W is the number of wavelengths of every fibre).Based on this, people such as Eramo propose wavelength shifter is designed to be the form (M.Listanti of each input optical fibre Channel Sharing, Wavelength converter sharing in aWDM optical packet switch:dimensioning and performance issues, ComputerNetworks, vol.32,2000:633-651), this structure proposes structure than people such as Danielsen and has almost saved 95% transducer, but the quantity of optical gate has certain increase.However, for the condition that the former supposed, promptly the number of wavelengths on every optical fiber is no less than 8, and the grouping that is 0.8Erlang for every fine load arrives, and still need utilize 8 transducers.
For non-burst service (random traffic), the FDL of same quantity is more effective than tunable wavelength transducer, and considerably cheaper, and therefore, for non-burst service, above-mentioned two kinds of schemes are compared with utilizing delay line to manage conflict, not economical and effective.Yet, for burst service, the business of middle-and-high-ranking burstiness particularly, the wavelength shifter of same quantity is more much effective than delay line.Therefore, we can be used in combination FDL and wavelength shifter, both can obtain lower cost, can solve the application of burst and non-burst service again simultaneously.
Summary of the invention:
The objective of the invention is at the deficiencies in the prior art, design proposes a kind of can effectively managing conflict, and the full optical packet switching node structure of cost, volume relatively low (little), this structure can be used for non-burst service, can support the burst service that day by day increases again.
If this switching fabric is supported N input optical fibre port and N output optical fibre port, every optical fiber is supported W wavelength channel (λ
1... λ
W), then the switching fabric of the present invention's proposition is made up of N 1 * W light wave decomposition multiplex device, a N W * 1 light wavelength division multiplexing, a R long transducer of tunable all-wave, M bar fibre delay line and one (WN+R+M) * (WN+R+M) clog-free optical switching matrix module.Wherein, M bar delay line has been formed the structure of degeneracy (degenerate) form, it is that continuous (step-length is T that duration that they postponed changes, T is the duration of a time slot), the cache-time that the delay line of shortest length provides grouping is T, and the cache-time that the longest delay line provides is MT.
Every input optical fibre links to each other with the input of a light wave decomposition multiplex device respectively, by demodulation multiplexer demultiplexing is carried out in the input grouping and handles, and the output of demodulation multiplexer connects the input port of space switching matrix, and input is grouped in carries out space switching in the switching matrix.Here, N bar input optical fibre needs N light wave decomposition multiplex device altogether, and each demodulation multiplexer has W delivery outlet, has taken WN input port of space switching matrix altogether.The output of space switching matrix links to each other with the input port of N light wavelength division multiplexing, and each multiplexer has taken W output port of switching matrix, and N multiplexer taken WN output port of space switching matrix altogether.The output of multiplexer connects output optical fibre.From annexation, input optical fibre, demodulation multiplexer, the space switching matrix, multiplexer, output optical fibre has been formed symmetrical structure.Space switching matrix remaining (R+M) fibre delay line and R long transducer of tunable all-wave to having connected M bar degeneracy form between input/output end port, the conflict that is used to divide into groups is handled.
Because be clog-free exchange fully, it is shared that R wavelength shifter and M bar fibre delay line can be entirely each input optical fibre, therefore, this project organization has following characteristics:
(1) grouping of coming out in the delay line can circular buffer, realizes bigger buffer memory ability; (2) owing to be shared structure, significantly reduce required delay line quantity and the wavelength shifter number of system, reduced the volume and the cost of switching equipment; (3) make full use of the wavelength-division multiplex technique of present maturation, can realize that multi-wavelength shares the load on the optical fiber, itself just to a certain degree alleviation packet collisions, therefore, when the number of wavelengths on the optical fiber increases, delay line that the present invention is used and wavelength shifter quantity and traditional design (are respectively each wavelength channel in input stage and dispose the long transducer of adjustable all-wave, or be each delivery channel configuration output state in output stage, or dual mode adopts simultaneously) maximum difference promptly is: not only used delay line and the transducer quantity of the present invention do not increase, but further reduce; (4) with in output stage compare for each output optical fibre port arrangement delay line, even each output optical fibre port only disposes a delay line (in fact being far above), this structure does not need more switching matrix port yet; (5) by wavelength-division multiplex technique back (wavelength is no less than 8) is shared in load, the present invention is for non-burst service, and what need only assurance delay line and transducer adds up to 8, and needn't make any regulation to the number of delay line and transducer.For moderate burst service, also if guarantee delay line and transducer add up to 8, and the quantity of wavelength shifter is 3～6 and gets final product.
Description of drawings:
Accompanying drawing 1 is a node structure schematic diagram of the present invention.
Whole switching fabric of the present invention comprises: input optical fibre (1), light wave decomposition multiplex device (2), clog-free space switching matrix (3), light wavelength division multiplexing (4), output optical fibre (5), the long transducer of tunable all-wave (6) and fibre delay line (7).
Accompanying drawing 2 utilizes delay line cache grouping, the schematic diagram that conflicts and solve for switching system.
Accompanying drawing 3 carries out wavelength conversion to the carrying wavelength of input grouping, the schematic diagram that conflicts and solve for switching system utilizes wavelength shifter.
When accompanying drawing 4 for the grouping of coming out the destination interface conflict takes place still from delay line, continue the schematic diagram of buffered packet to manage conflict.
When the destination interface conflict still takes place for the grouping of coming out in accompanying drawing 5 from delay line, by the schematic diagram of wavelength conversion to manage conflict.
Accompanying drawing 9 is for the present invention is directed to burst service (load is 0.8/ every fibre), and the burst packets average length is 10 grouping time slots.The wavelength shifter of varying number and the combining and configuring of delay line are to the schematic diagram that influences of discarded ratio of packet.Wherein, R is a transducer quantity, and M is a delay line quantity, and the input-output optical fiber number is N=16.
Embodiment:
Be described in detail below in conjunction with the embodiment of accompanying drawing technical solution of the present invention.
Fig. 1 is a switching node structural representation of the present invention.If switching fabric of the present invention is supported N input optical fibre port and N output optical fibre port, every optical fiber can be supported W wavelength channel, is designated as { λ
1... λ
W, whole switching fabric then of the present invention is by the individual 1 * W light wave decomposition multiplex device of N, a N W * 1 light wavelength division multiplexing, (WN+R+M) * (WN+R+M) clog-free optical switching matrix, R the long transducer of tunable all-wave, M bar fibre delay line composition.Wherein, M bar fibre delay line has been formed the structure of degeneracy form, be that the duration variation that this M bar fibre delay line is postponed is continuous, step-length is T, T is the duration of a time slot, the cache-time that the delay line of shortest length provides grouping is T, and the cache-time that the longest delay line provides is MT.
Every input optical fibre (1) links to each other with the input of a light wave decomposition multiplex device (2) respectively, by demodulation multiplexer (2) demultiplexing being carried out in the input grouping handles, the output of demodulation multiplexer connects the input port of space switching matrix (3), and input is grouped in carries out space switching in the switching matrix.Here, N bar input optical fibre needs N light wave decomposition multiplex device (2) altogether, and each demodulation multiplexer has W delivery outlet, has taken WN input port of space switching matrix altogether.The output of space switching matrix (3) links to each other with the input port of N light wavelength division multiplexing (4), and each multiplexer (4) has taken W output port of switching matrix, and N multiplexer taken WN output port of space switching matrix altogether.The output of N multiplexer (4) connects N bar output optical fibre (5).Input optical fibre (1), demodulation multiplexer (2), space switching matrix (3), multiplexer (4), output optical fibre (5) has been formed symmetrical structure.Space switching matrix (3) remaining (R+M) fibre delay line and R long transducer of tunable all-wave to having connected M bar degeneracy form between input/output end port, the conflict that is used to divide into groups is handled, the input/output end port that is the NW+1 to NW+M of space switching matrix (3) is respectively applied for the fibre delay line that connects M bar degeneracy form, and the input/output end port of NW+M+1 to NW+M+R is respectively applied for and connects R the long transducer of tunable all-wave.
Because be clog-free exchange fully, it is shared that R wavelength shifter and M bar fibre delay line can be entirely each input optical fibre.
Behind the grouping process demodulation multiplexer demultiplexing that arrives in the input optical fibre, enter switching matrix.Grouping in input (output) channel can be by set of variables I (i, λ
j) (O (i, λ
j)) be described, wherein, i{i=1 ..., N} represents input (output) optical fiber, λ
jJ=1 ..., W} represents wavelength.Grouping has regular length, and the grouping on each wavelength channel arrives and to be based on slot synchronization, and a time slot has been represented the single transmission packets time, time quantum of the duration of a time slot (T) expression.Order:
When being located at the t time slot, have I (1, λ
1, t)=1 and I (N, λ
1, t)=1, and their destination interface is first output optical fibre.If exchange control scheduler permission I (1, λ
1) directly enter O (1, λ
1), promptly [I (1, λ
1, t) → and O (1, λ
1, t)]=1, then for I (N, λ
1), below any one situation can manage conflict when satisfying:
(1) if there is λ
jJ=2 ..., W} and k ∈ 1 ..., R}, satisfy O (1, λ
j, t)=0 and WC (k, t)=0;
(2) if exist k ∈ 1 ..., M}, satisfy FDL (k, t)=0;
First kind of situation shows, can utilize wavelength shifter, and in the output optical fibre on the idle wavelength, this situation as shown in Figure 3 with the carrying wavelength Conversion of conflict grouping; Second kind of situation then shows, can utilize delay line, the packet scheduling of conflict is transmitted to time slot in the future, and this situation is (dotted line among the figure shows that the buffer memory that is grouped in that is buffered just enters output port 1 afterwards, has staggered with the grouping that conflicts in time) as shown in Figure 2.When either way satisfying, adopt wavelength shifter with manage conflict (favourable) to reducing time delay.
It can also be seen that by the connected mode in the accompanying drawing 1, when the purpose conflict still takes place in the grouping of coming out from delay line, also can continue this grouping cache is maybe carried out this grouping wavelength conversion with manage conflict (when either way satisfying, preferentially adopting wavelength shifter to manage conflict).(among Fig. 4 Fig. 5, the carrying wavelength of the grouping of coming out from delay line is λ to both of these case shown in accompanying drawing 4 and accompanying drawing 5 respectively
1, destination interface is the 1st output optical fibre).
From the discarded ratio of packet result of accompanying drawing 6～9 as can be seen: for non-burst service and the lower burst service of packeting average length, delay line more can effectively manage conflict than transducer (1).When the number of wavelengths of every fibre is no less than 8, as long as the total quantity of delay line and transducer reaches 8, needn't make any regulation to the concrete quantity configuration of delay line and transducer, can guarantee that grouping abandons performance; (2) for the burst service (5＜l＜10) of medium burstiness, along with increasing of transducer quantity, discarded ratio of packet will reduce gradually, and promptly transducer is more effective than delay line.If but disposed 6 transducers, 2 delay lines can be supported burst service preferably; (3) for different service environments, load size and I/O fiber port number, the transducer of sharing and the concrete quantity of delay line understand that some is different, but howsoever, among the present invention delay line and transducer are designed to cost and the volume that the shared structure of feedback will significantly reduce switching system simultaneously.
Claims (1)
1, a kind ofly support to happen suddenly and the full optical packet switching node structure of non-burst service, it is characterized in that by N 1 * W light wave decomposition multiplex device, N W * 1 light wavelength division multiplexing, one (WN+R+M) * (WN+R+M) clog-free complete optical switching matrix module, R the long transducer of all-wave, M bar fibre delay line is formed symmetrical structure, wherein, N is an I/O fiber port number, W can transmit the wavelength channel number for each port, N bar input optical fibre links to each other with the input of N light wave decomposition multiplex device respectively, the NW of a N demodulation multiplexer output links to each other with WN input of space switching matrix, the WN of a space switching matrix output port links to each other with the input port of N light wavelength division multiplexing, the output of N multiplexer connects N bar output optical fibre, space switching matrix remaining (R+M) fibre delay line and R long transducer of tunable all-wave to having connected M root degeneracy form between the input/output end port, the cache-time that the delay line of shortest length provides grouping is the duration T of a time slot, the cache-time that the longest delay line provides is MT, long transducer of all-wave and fibre delay line are shared form for feedback simultaneously, delay line and the long transducer of all-wave add up to 8, wherein the quantity of the long transducer of all-wave is 3～6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031295371A CN1235357C (en) | 2003-06-26 | 2003-06-26 | Full optical packet switching node structure for supporting burst or non-burst businesses |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031295371A CN1235357C (en) | 2003-06-26 | 2003-06-26 | Full optical packet switching node structure for supporting burst or non-burst businesses |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1472969A CN1472969A (en) | 2004-02-04 |
| CN1235357C true CN1235357C (en) | 2006-01-04 |
Family
ID=34153574
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB031295371A Expired - Fee Related CN1235357C (en) | 2003-06-26 | 2003-06-26 | Full optical packet switching node structure for supporting burst or non-burst businesses |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1235357C (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100346619C (en) * | 2004-07-02 | 2007-10-31 | 电子科技大学 | Light burst/grouping switching structure without team head block |
| CN101193049B (en) * | 2006-11-24 | 2010-09-29 | 中兴通讯股份有限公司 | A core node switching structure for integrated edge node |
| CN101193050B (en) * | 2006-11-24 | 2011-12-28 | 中兴通讯股份有限公司 | A method for data receiving and transmitting of core node switching device in optical sudden network |
| CN101212818B (en) | 2006-12-26 | 2010-12-08 | 华为技术有限公司 | Network node and buffer device, and scheduling method |
| CN101079680B (en) * | 2007-05-25 | 2011-03-30 | 北京邮电大学 | Full optical serial multicast module based on multi-level buffer share and its multicast method |
| CN101742411B (en) * | 2008-11-25 | 2016-08-03 | 中兴通讯股份有限公司 | A kind of method of dynamic adjustment paging capability |
| CN102075821B (en) * | 2010-12-27 | 2014-04-16 | 重庆邮电大学 | Shared wavelength converter device and optical packet collision solving method |
| CN102916890B (en) * | 2012-10-17 | 2015-05-27 | 重庆邮电大学 | Shared wavelength conversion device and method for solving multicast competition |
| CN105721097A (en) * | 2016-01-25 | 2016-06-29 | 南京大学 | High-speed long-distance switching matrix system for broadband microwave signal |
| CN107071594B (en) * | 2017-05-05 | 2020-03-24 | 重庆邮电大学 | Multicast switching node conflict resolution method based on spectrum conversion and optical cache technology |
| CN111615022B (en) * | 2020-05-20 | 2022-04-19 | 重庆邮电大学 | Conflict resolution method for all-optical nodes of space division multiplexing elastic optical network |
-
2003
- 2003-06-26 CN CNB031295371A patent/CN1235357C/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN1472969A (en) | 2004-02-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hunter et al. | Approaches to optical Internet packet switching | |
| KR101978191B1 (en) | Scalable optical switches and switching modules | |
| Danielsen et al. | Optical packet switched network layer without optical buffers | |
| US5878177A (en) | Layered switch architectures for high-capacity optical transport networks | |
| CN1147327A (en) | Optical telecommunications network | |
| JP2002185482A (en) | Composite packet switching method and system by transmissive photonic slot routing using wdm | |
| CN1332546A (en) | Light IP exchange route structure | |
| CN1235357C (en) | Full optical packet switching node structure for supporting burst or non-burst businesses | |
| JPH11252126A (en) | Large capacity wavelength division multiplex optical atm switch | |
| CN1489323A (en) | Wave-division multiplex optical exchanging system | |
| Papadimitriou et al. | Multiwavelength Optical LANs | |
| CN101370315A (en) | Sharing type optical packet switching node structure based on inside wavelength | |
| CN1324830C (en) | Expandable multicasting light exchange structure with light-regulating shunt | |
| EP1367754A2 (en) | WDM metropolitan access network architecture based on hybrid switching | |
| CN1642061A (en) | Optical burst exchange node with internal acceleration | |
| CN1162998C (en) | Multifnctional light split/insersion multiplexer | |
| EP2168395B1 (en) | Maintaining correlated virtual data streams through a network | |
| Shaddad et al. | Contention resolution of optical burst switching for data center | |
| Han et al. | Design of AWG-based WDM-PON architecture with multicast capability | |
| Pattavina | Multi-wavelength switching in IP optical nodes adopting different buffering strategies | |
| CN1295394A (en) | Optical cross connection node for crossing between fiber link and wavelength groups | |
| Vinolee et al. | Conversion complexity of multicast routing and wavelength assignment converters with different wavelength conversion in Benes network | |
| CN1283058C (en) | Structure of optical switching node based on sharing convertor of adjustable wavelength | |
| Pedro et al. | Towards fully flexible optical node architectures: Impact on blocking performance of dwdm transport networks | |
| CN2914114Y (en) | Optical module on optical wavelength division multiplex device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C19 | Lapse of patent right due to non-payment of the annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |