CN101540933B - Programmable dynamic bandwidth distribution method and device - Google Patents

Abstract

The invention discloses a programmable dynamic bandwidth distribution method and a device, which has the advantages of the prior art; a programming software at an OLT end can freely configure the performance of an access network instantly by re-defining parameters, and leads programming software to be a really open system environment by a high-performance ARM-series CPU embedded in an ONU/OLT. According to the parameters of the bandwidth of the service, the network time delay, user priority, and the like, a PON system is optimized by aiming at special requirements. Simultaneously, the device has programmability and strong flexibility, the bandwidth distribution arithmetic can be adjusted without replacing the network device, the network user priority can be re-classified and the intercommunication between different systems and non-compatible equipment are realized.

Description

Programmable dynamic bandwidth distribution method and device

Technical field

The invention belongs to the optical communication technique field, relate to PON (EPON) communication, more specifically, the present invention relates to the Programmable Dynamic allocated bandwidth device and method of EPON.

Background technology

EPON (ethernet passive optical network) adopts the ethernet data link layer agreement of expansion on the architectural framework basis of EPON, have advantages such as agreement is simple, cost is low, professional ability is strong.Since the EPON notion proposes, received showing great attention to of industry.The EFMA of IEEE (institute of electrical and electronic engineers) (first kilometer Ethernet alliance) and the technical specification IEEE 802.3ah that has issued EPON in June, 2004 have received the extensive support of industry, and relevant chip and equipment are all released, and certain application is also arranged.

Experience from xDSL (x DSL) technology and mobile communication technology development; The opening that realizes the EPON network terminal can effectively reduce network construction cost and O&M cost; Promoting it to use, help forming complete industrial environment, is the trend of technical development.Can therefore, realize the intercommunication of OLT of all big enterprises (optical line terminal) and ONU (optical network unit), be that can measurement EPON technology ripe key.Currently come, the OLT of all big enterprises and ONU (the particularly OLT of each large chip manufacturer and ONU) intercommunication is still difficult to be realized.Have its source in: at first be IEEE 802.3ah agreement, the option of identical concept is more, like transmission range, FEC (forward error correction) etc. different selections is arranged; Management and business also respectively are standard; Secondly, different chip manufacturer, the different system view, the method that agreement realizes is different; Define different privately owned Extended Protocols again, caused chip and equipment compatibility problem.

It seems from the reason that influences the EPON interoperability, mainly is the intercommunication of three levels: transport layer, operation layer and management level, the wherein most importantly intercommunication between the key technology DBA of transport layer (Dynamic Bandwidth Allocation) algorithm.

In practical application, because each equipment vendors is different to the definition of DBA, even therefore with a kind of algorithm, it has implementation method and also can vary.Simultaneously, existing bandwidth allocation algorithm all is to be embedded in optic communication device the inside, because of its be solidify and real-time change poor, adjust so be difficult to the part of implementation algorithm.And Network Provider can not realize and adjust the allocated bandwidth logic of oneself after network design is good.Cause,, must change device or redeploy network topology if when existing algorithm and assignment logic adjusted; Not only cause the significant wastage of resource; Reduce the profit of Network Provider, and influenced user's normal use, caused unnecessary loss.So; Operators press for a kind of programmable software of definition now; The DBA algorithm able to programme that can adjust at any time according to own demand, generally these definition comprise the division to the grade of service, the requirement of time delay, the distribution mechanism of remaining bandwidth etc., thereby realize the purpose that interconnects.

Summary of the invention

In order to overcome above-mentioned shortcoming, the technical problem that the present invention solved just provides a kind of Programmable Dynamic allocated bandwidth device and method, and this method realizes to the making full use of of network bandwidth resources, to repartitioning of user gradation to greatest extent.

Technical scheme of the present invention is: a kind of programmable dynamic bandwidth distribution method; Concrete steps: (a) local device powers on; ARM CPU calls a bandwidth allocation algorithm to programmable module, initialization various piece parameter and user's running state parameter from storage and ancillary equipment; (b) when the PON of this locality Access Network and other networks communicate, the OLT of local PON detects the type of the specified bytes of accepting bag, the network that judgement will be communicated by letter whether with local PON system compatible; (c) if compatible, then continue to use the algorithm communication in the current programmable module; (d) if incompatible, then notify ARM series CPU, whether ARM CPU at first detection of stored and ancillary equipment has the algorithm of suitable current communication, if having, then algorithm is loaded into programmable module, continues current communication; Otherwise, through the algorithm of serial port chip or storage and the suitable current communication of ancillary equipment loading; (e) some ONU need repartition classification in network, and when needing to promote the rank of oneself, the ARM CPU of ONU sends to OLT and asks for instructions; (f) the ARM CPU of OLT issues ONU with other respective algorithms of correlation level, and the CPU of ONU is loaded into corresponding programming module or memory device with it; (g) when the user of ONU wants to abandon or promote the rank of oneself, the ARM CPU of ONU readjusts the user's bandwidth assignment logic.

A kind of Programmable Dynamic allocated bandwidth device, the ONU of this device comprise PHY chip, ARM series CPU, programmable module, serial port chip, MAC chip, storage and ancillary equipment; Programmable module is direct-connected with PHY chip, MAC chip, ARM CPU, serial port chip respectively, and direct-connected another PHY chip user of MAC chip, an also direct-connected storage and the ancillary equipment of ARM CPU; ARM CPU adopts identical clock signal with serial port chip, and two PHY chips adopt another identical clock signal with MAC chip and programmable module; OLT comprises PHY chip, ARM series CPU, programmable module, serial port chip, MAC chip, storage and ancillary equipment, multiport PHY chip, SWIYCH chip; Its internal structure relation is similar with ONU, replaced MAC chip and coupled PHY chip among the ONU with multiport PHY chip and SWITCH chip respectively; The ARM CPU of OLT adopts the clock signal identical with the ARM CPU of ONU with serial port chip; Programmable module, multiport PHY chip, SWITCH also adopt the clock signal identical with the PHY of ONU; The clock signal of the PHY chip of OLT will have the identical clock signal that communicates with network; This device characteristic is: OLT, ONU end have embedded ARM series CPU, have added programmable module at OLT, ONU end simultaneously; When ARM series CPU made OLT, ONU become real development environment, the correlation function that can effectively hold OLT, ONU changed the demand with the variation user who adapts to network in real time, and interconnecting between different PON system.Integrated DBA algorithm has embedded ARM series CPU simultaneously in the programmable module of ONU and OLT end, may operate on the standard operation system.The performance of ARM series CPU will make ONU become real development system environment, simultaneously can also the standardized third party software of compatibility.

Be to guarantee under the constant prerequisite of existing network function and structure, realize to greatest extent to the making full use of of network bandwidth resources, to repartitioning of user gradation.Simultaneously, device of the present invention (like Fig. 1, shown in 2) has embedded high-performance ARM series CPU, can run on standard operation system, like Linux and VxWorks.This performance makes it become real open system environment, and the compatibility standard third party software, as is used for exchange, VOIP (networking telephone) and embedded web server function.Programmable dynamic bandwidth allocation algorithm provides advanced flow report and management function.In the frequent flow rate mode that changes of expection and the application in future, support high quality-of-service (QoS) and high network efficiency.Have only the smart that is integrated with the DBA algorithm of advanced programmable through use, just possibly tackle fast changing network, so that the current and following service better is provided, interconnecting between EPON just can guarantee to realize.

ARM series CPU and ONU module able to programme subpackage engine flexibly can be accomplished senior mobile classification, realize the separation through the network service, make service secure.This programmable scheme can be directed against allocated bandwidth simultaneously, postpones to optimize or the definition of the grade of service newly, for the service supplier provides more precise information to formulate optimum service plan.According to service that provides and residing competitive environment, the service supplier has different demands when user network bandwidth and traffic delay are provided.Through using DBA algorithm able to programme flexibly, operator can develop the service plan of customization according to the final service that provides and bandwidth, delay requirement (IPTV (IPTV), VOIP, game on line etc.).For example: queue management mechanism and equity dispatching mechanism, and fully the utilized bandwidth resource (like Fig. 3: (1) (2) (3)) eliminate mechanism.Simultaneously, this algorithm can pass through related software, and it is carried out the operation able to programme of Denging, loads corresponding equity dispatching mechanism (Fig. 5), utilized bandwidth is eliminated mechanism (Fig. 6,7) etc.Therefore can solve a difficult problem of present EPON, the intercommunication between different system.

The invention has the advantages that: the present invention includes classical bandwidth allocation algorithm and can be easy to existing all-network compatible; The present invention can not changing device or redeploying under the network topology prerequisite, adjust existed algorithms and assignment logic after network design is good; The present invention can make the programmable software of operator through the terminal, realizes the adjustment of distribution mechanism of requirement, the remaining bandwidth of division, time delay optimization to the grade of service; The present invention can be at different PON inter-net communications, thus the supplier who has realized heterogeneous networks, the purpose that interconnects between incompatible device systems.OLT and ONU end the is integrated programmable module of ARM series CPU can be supported all DBA algorithms, can be according to the definition of allocated bandwidth, time delay or the new grade of service etc., advanced flow report and related management function is provided.This flexibility DBA able to programme makes the service supplier after network design, still can determine to realize and adjust the allocated bandwidth logic of oneself, for the service supplier provides more precise information to formulate optimum service plan.

Description of drawings

Fig. 1 is the ONU structural representation of Programmable Dynamic allocated bandwidth device of the present invention; 1:PHY (physical layer) chip; 2: storage and ancillary equipment; 3: serial port chip; 4: programmable module; 5:ARM series CPU; 6:MAC (medium access control sublayer agreement) chip; 7: user's Ethernet data; 8: clock signal; 9: optical interface; The 10:RS232 serial ports; 11:MII (GMII).

Fig. 2 is the OLT structural representation of Programmable Dynamic allocated bandwidth device of the present invention; The 1:PHY chip; 2: storage and ancillary equipment; 3: serial port chip; 4: programmable module; 5:ARM9CPU; 6:RMII (RMII); 7:SWITCH (switch); 8: clock signal; 9: optical interface; The 10:RS232 serial ports; 11:MII; 12: Ethernet data; 13: multiport PHY chip; 14 Ethernet interfaces.

Fig. 3 is the inabundant utilized bandwidth resource sketch map of present bandwidth allocation methods;

Fig. 4 loads flow chart for programmable dynamic bandwidth distribution method algorithm of the present invention;

Fig. 5 is the flow chart of programmable dynamic bandwidth distribution method fair algorithm of the present invention;

Fig. 6 does not eliminate flow chart for programmable dynamic bandwidth distribution method of the present invention uses time slot;

Fig. 7 is a not utilized bandwidth elimination flow chart of programmable dynamic bandwidth distribution method of the present invention.

Embodiment

Below in conjunction with concrete figure the present invention is further described:

Fig. 1,2 is respectively the structure of the ONU and the OLT of Programmable Dynamic bandwidth allocation algorithm, and core is ARM9 CPU 5, and this paper is example with high performance ARM9 CUP, and this performance makes it become real development system environment.Wherein programmable module 4 is the bandwidth allocation algorithms that are used for storing in the current PON system, and storage and ancillary equipment 2 are the allocation algorithms that are used for storing current classics.When the algorithm in the programmable module 4 was not suitable for current communication, ARM9 CPU 5 just at first sought whether suitable algorithm is arranged in storage and ancillary equipment 2, if having, call in programmable module 4 again, continues communication; Otherwise serial port chip 3 or storage and ancillary equipment 2 through OLT load the DBA algorithm that is fit to current communication.Load related algorithm by the ARM9 CPU 5 of the 5 notice ONU of OLT and to ONU then, make this system adapt to current communication rapidly,, guarantee user's proper communication so that keep the unimpeded of communication.The loading flow process is as shown in Figure 4.

Fig. 5 is the flow chart of equity dispatching mechanism: adopt the principle of first in first out, after one-period finished, next cycle could send, and total data packet number of transmission has the decision of mandate.In the m cycle, refresh T _Ij ^S(m), make it point to the buffers packet T that arrives the earliest in all formations _Ij ^S(m).Before scheduling buffer memory bag, all to check T _Ij ^F-L ^PackageWhether＜0 set up: if be false, also do not use up in this formation authorized window of explanation, packet is sent out, and refreshes T simultaneously _Ij ^S(m), point in the formation the next time of advent of buffers packet the earliest.If cycle this moment does not finish, then refresh T _Ij ^S(m) all have the formation of zero-time buffers packet the earliest to make it point to ONU, and above process then circulates.If set up, then explain should business in this cycle authorized window use up, can not send data, then with T _Ij ^S(m) be set to ∞.Because the randomness of packet possibly remain in addition in the service class, but be not enough to send next packet in formation, therefore just produced untapped formation residue and sent length UPR _Ij, the UPR of all formations _IjAnd do ${\mathrm{UPR}}_i=\underset{j}{\overset{J}{\mathrm{\Σ}}}{\mathrm{UPR}}_{\mathrm{ij}}.$ The idiographic flow of this fair algorithm is as shown in Figure 5.

Fig. 6 eliminates flow chart for using time slot: 1) when total request of all ONU during less than maximized window, authorize for each ONU according to demand, if for the big or small t of each ONU authorized window _i ^g(m)=t _r(m), then according to the demand L of each formation _Ij ^R(m) come to authorize to them.2) if when total request of all ONU during greater than maximized window, then distribute bandwidth according to consecutive mean SLA (service-level agreement) weight of ONU, wherein consecutive mean SLA weight is: $t_i^{\mathrm{SLA}}\left(m\right)=\frac{{\mathrm{\β}}_i\left(m\right)R_i\left(m\right)}{\underset{i}{\overset{I}{\mathrm{\Σ}}}\left({\mathrm{\β}}_l\left(m\right)R_i\left(m\right)\right)}(T_{\mathrm{Cycle}}^{\mathrm{MAX}}-t^p-I\×t_g-I\×t_r)$ , to authorize according to the SLA queue length for each ONU is also corresponding, the SLA queue length is: $L_{\mathrm{Ij}}^{\mathrm{SLA}}\left(m\right)=\frac{{\mathrm{\β}}_{\mathrm{Ij}}\left(m\right)L_{\mathrm{Ij}}^R\left(m\right)C}{\underset{k}{\overset{J}{\mathrm{\Σ}}}\left({\mathrm{\β}}_{\mathrm{Ik}}\left(m\right)L_{\mathrm{Ik}}^R\left(m\right)\right)}{t}_i^g\left(m\right).$

If the t of an ONU _i ^SLA(m) greater than R _i(m), t then _i ^g(m) equal R _i(m), calculate the residue window value simultaneously, the residue window of all ONU with: $\&dtri;t=\underset{k}{\overset{I}{\mathrm{\&Sigma;}}}\mathrm{Max}(L_{\mathrm{Ik}}^{\mathrm{SLA}}\left(m\right)-L_{\mathrm{Ij}}^R,0),$ Then it is distributed to the ONU that can not satisfy condition: $N_a\left(m\right)=\left\{k\right|\mathrm{Min}(t_k^{\mathrm{SLA}}\left(m\right)-R_k\left(m\right)/C,0)<0\}.$ Distribute and still adopt the SLA authorization method: $t_i^{\mathrm{SLA}}\left(m\right)=t_i^{\mathrm{SLA}}\left(m\right)+\frac{{\mathrm{\&beta;}}_i\left(m\right)R_i\left(m\right)}{\underset{k}{\overset{N_a\left(m\right)}{\mathrm{\&Sigma;}}}\left({\mathrm{\&beta;}}_k\left(m\right)R_k\left(m\right)\right)}\&dtri;t.$ For each service class formation, if $L_{\mathrm{Ij}}^{\mathrm{SLA}}\left(m\right)>L_{\mathrm{Ij}}^R\left(m\right),$ Then $L_{\mathrm{Ij}}^{\mathrm{SLA}}\left(m\right)=L_{\mathrm{Ij}}^R\left(m\right)$ , calculate the length of residual queue simultaneously, all residual queue's length sums are: $\&dtri;t=\underset{k}{\overset{J}{\mathrm{\&Sigma;}}}\mathrm{Max}(L_{\mathrm{Ik}}^{\mathrm{SLA}}\left(m\right)-L_{\mathrm{Ij}}^R,0)$ , it is distributed to the formation that all demands fail to be satisfied: $N_a\left(m\right)=\left\{k\right|\mathrm{Min}(L_{\mathrm{Ik}}^{\mathrm{SLA}}\left(m\right)-L_{\mathrm{Ik}}^R,0)<0\}$ , distribute equally also according to the SLA queue length: $L_{\mathrm{Ij}}^{\mathrm{SLA}}\left(m\right)=L_{\mathrm{Ij}}^{\mathrm{SLA}}\left(m\right)+\frac{{\mathrm{\&beta;}}_{\mathrm{Ij}}\left(m\right)L_{\mathrm{Ij}}^R\left(m\right)}{\underset{k}{\overset{N_a\left(m\right)}{\mathrm{\&Sigma;}}}{\mathrm{\&beta;}}_{\mathrm{Ik}}\left(m\right)L_{\mathrm{Ik}}^R\left(m\right)}\&dtri;t$ , this cyclic process is performed until quantity and the formation of not satisfying the ONU that regulates and no longer changes, and authorized window then is described and is authorized formation by whole utilizations, then loop ends.

Fig. 7 eliminates flow chart for using time slot, is divided into: 1) own elimination 2 in this ONU) eliminate between ONU.

1) eliminates in the ONU

In any ONU all formations residues send length with ${\mathrm{UPR}}_i=\underset{j}{\overset{J}{\mathrm{\&Sigma;}}}{\mathrm{UPR}}_{\mathrm{Ij}}$ , it maybe be greater than the data packet length in a certain formation, so, refresh T again _Ij ^S(m) make its point to the packet that reaches the earliest of current queue after finishing scheduling, refresh T simultaneously _Ij ^S(m) it is pointed to and contain the formation that arrives packet the earliest, judge then $T_{\mathrm{Ij}}^F\left(m\right)\&GreaterEqual;L^{\mathrm{Package}}$ Whether set up,, explain that then authorized window is not enough, can not send packet, then be provided with if be false $T_{\mathrm{Ij}}^S\left(m\right)=\&infin;$ If set up, then explanation can also continue to send packet, judges UPR _iWhether less than the length of data package l that will send _Ijp(t _Ijp), if less than data packet length, then explanation can not be sent the packet setting $T_{\mathrm{Ij}}^S\left(m\right)=\&infin;$ ,, reduce UPR simultaneously if redispatch greater than data packet length _i=UPR _i-l _Ijp(t _Ijp), if also have packet then to continue said process.

2) eliminate between ONU

When UPR can not be eliminated in ONU inside _iThe time, the UPR of then all ONU _i, just need to eliminate between ONU, set up a question blank at the OLT end, the USR=UPR of an overall situation of definition _i, be q of each ONU definition _i=t _i(m)+t _g+ t _r-RTT _i, and according to q _iValue descending ONU is sorted, at this moment, satisfy interface condition between adjacent ONU: as an ONU data window r _i(m)+t _g+ t _rWhen very big or RTT _iWhen very little.Propose first ONU as the person of handing the baton in the question blank of OLT after the ordering, on its authorized window, increase the residual UPR of last ONU _i=0, equal demand if satisfy the demand of this ONU then authorize this moment, and from new calculating $\mathrm{UPR}=t_i^g-R_i\left(m\right)$ , begin to authorize to make the ONU send window opening time shift to an earlier date UPR _iTime.Then, propose second ONU, judge whether to satisfy t as anchor man _i(m)+t _g+ t _r-RTT _i/ 2>=RTT _I+1/ 2, if do not satisfy then according to normal mechanism second ONU authorized, UPR is wasted, otherwise OLT waits until and authorize just for second ONU after the relay baton of first ONU arrives, simultaneously it as the person of handing the baton, the repetition said process is up to end cycle.

Claims (3)

1. programmable dynamic bandwidth distribution method, concrete steps: (a) local device powers on, and ARM CPU calls a bandwidth allocation algorithm to programmable module, initialization various piece parameter and user's running state parameter from storage and ancillary equipment; (b) when the passive optical network PON Access Network of this locality and other networks communicated, the optical line terminal OLT of local PON detected the type of the specified bytes of accepting bag, the network that judgement will be communicated by letter whether with local PON system compatible; (c) if compatible, then continue to use the algorithm communication in the current programmable module; (d) if incompatible, then notify ARM series CPUARM CPU, whether ARMCPU at first detection of stored and ancillary equipment has the algorithm of suitable current communication, if having then algorithm is loaded into programmable module, continues current communication; Otherwise, through the algorithm of serial port chip or storage and the suitable current communication of ancillary equipment loading; (e) some optical network unit ONU need be repartitioned classification in network, and when needing to promote the rank of oneself, the ARM CPU of ONU sends to OLT and asks for instructions; (f) ARMCPU of OLT issues ONU with other respective algorithms of correlation level, and the ARM CPU of the CPUONU of ONU is loaded into corresponding programming module or memory device with it; (g) when the user of ONU wants to abandon or promote the rank of oneself, the ARM CPU of ONU readjusts the user's bandwidth assignment logic.

2. according to the said programmable dynamic bandwidth distribution method of claim 1, it is characterized in that: integrated Dynamic Bandwidth Allocation DBA algorithm has embedded ARM series CPUARM CPU simultaneously in the programmable module of ONU and OLT end, may operate on the standard operation system.

3. according to claim 2 and 1 said programmable dynamic bandwidth distribution method, it is characterized in that: the performance of ARM series CPUARM CPU will make ONU become real development system environment, simultaneously can also the standardized third party software of compatibility.

Images