Fast optical receiver recovery on the client side using optical interconnect with DSP

Abstract

The present disclosure relates to a system and method for fast digital signal processor (DSP) optical receiver recovery, i.e., for use in an optical modem configured on the client side. This method can be used in optical protection switch applications to allow fast switching between two client links, i.e., within 50 ms. A receiver (Rx) digital signal processor (DSP) (20) in the optical receiver includes a device configured to detect (102) a traffic interruption on the current link, enter (104) a delay period, and perform one of the following (106): i) receive good traffic during the delay period; ii) expire the delay period, thereby notifying the host device and retraining to acquire an optical signal.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] The present disclosure claims priority to U.S. Provisional Patent Application No. 63 / 272,133, filed on October 26, 2021, entitled “Fast Optical Receiver Recovery on the Client Side Using Optical Interconnects with DSP,” the contents of which are incorporated herein by reference in their entirety. Technical Field

[0003] The present disclosure relates generally to optical networks. More specifically, the present disclosure relates to a system and method for fast digital signal processor (DSP) optical receiver recovery, ie, for optical interconnects deployed at the client side. Background Art

[0004] Optical modems (transceivers, transmitters and receivers, pluggable modules, etc.) are equipped with DSP circuits to perform various functions. Optical interconnects can include non-return to zero (NRZ), pulse amplitude modulation (PAM), and coherent modulation (e.g., quadrature amplitude modulation (QAM), etc.). It is worth noting that DSP can be used for any such modem. The optical link is formed by a transmitter (Tx) connected to a receiver (Rx) via an optical fiber interconnect. There are a variety of applications that require optical fiber switches for protection, such as optical protection switches (OPS), and this requires that Tx be quickly switched to a different optical fiber, and Rx be configured to receive Tx from a different optical fiber. For example, the traditional requirement is that the time required for Tx and Rx to get back online is 50 milliseconds or less. However, in a typical DSP implementation, the recovery between Tx and Rx takes several seconds, or even up to 10s or more.

[0005] On the line side, efforts to reduce this time were made in commonly assigned U.S. Patent No. 9,100,116, issued on August 4, 2015, entitled "Short-Term Optical Recovery System and Method for Coherent Optical Receivers," the contents of which are incorporated herein by reference in their entirety. The approach focuses primarily on improving the recovery speed in the line-side coherent modem. Here, the switches are typically located on different fibers that may have very different parameters (e.g., distance, fiber type, dispersion, loss, etc.). Summary of the invention

[0006] The present disclosure relates to systems and methods for fast digital signal processor (DSP) optical receiver recovery, i.e., for optical interconnects configured at the client side. A receiver (Rx) digital signal processor (DSP) in the optical receiver includes a device configured to detect a traffic interruption on the current link, enter a delay period, and perform one of the following: i) receive good traffic during the delay period; ii) expire the delay period, thereby notifying the host device and retraining to acquire optical signals.

[0007] In one embodiment, a receiver (Rx) digital signal processor (DSP) in an optical receiver is described, and the Rx DSP includes a signal processor configured to detect a traffic interruption on a current link associated with the optical receiver, enter a delay period, and perform one of the following: i) receive good traffic during the delay period; ii) expire the delay period to notify a host device and retrain to acquire an optical signal.

[0008] The circuit can be configured to avoid retraining to acquire the optical signal during the delay period and avoid any notification to the host device. The circuit can be configured to change the setting of the optical signal to one of the default setting or predetermined setting of the new link after entering the delay period. During the delay period, the setting for the optical signal can be maintained. The notification includes one or more of loss of signal (LOS) and loss of light (LOL). The circuit can also be configured to receive a configuration used during the delay period.

[0009] The circuit can also be configured to, in response to receiving good traffic within the delay period, continue to process good traffic and retrain the optical signal while processing good traffic. The optical receiver can be connected to an optical protection switch (OPS). The optical receiver can be connected to an optical coupler, where two different links including the current link can send optical signals. The corresponding optical transmitters on the two different links can be configured for fast closing and opening.

[0010] The optical signal can include non-return-to-zero (NRZ) modulation, pulse amplitude modulation (PAM), or coherent modulation.The transmitter of the current link and the transmitter of the protection link can have settings adjusted so that the RxDSP does not need to be adjusted when switching between links.

[0011] In another embodiment, a method implemented in a receiver (Rx) digital signal processor (DSP) in an optical receiver includes detecting a traffic interruption on a current link associated with the optical receiver, entering a delay period, and performing one of: i) receiving good traffic during the delay period; ii) expiring the delay period, thereby notifying a host device and retraining to acquire an optical signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The present disclosure is illustrated and described herein with reference to the various drawings, wherein like reference numerals are used to refer to like system components / method steps where appropriate, and wherein:

[0013] Figure 1 It is the block diagram of OPS module.

[0014] Figure 2 It is the OPS block diagram of the Rx side modem.

[0015] Figure 3 is the block diagram of the TPT module.

[0016] Figure 4 This is a block diagram of TPT on the Rx side.

[0017] Figure 5A is a TPT configuration diagram showing the laser off / on and Rx recovery requirements. Figure 5B is based on Figure 5A Optical Rx characteristic diagram of the switch obtained with the configuration.

[0018] Figure 6 is a diagram showing three methods of DSP Rx fast recovery.

[0019] Figure 7 is a diagram of an example of TPT and OPS protection.

[0020] Figure 8 is a flow chart of a process that may be implemented by a circuit such as an Rx DSP.

[0021] Fig. 9A is a diagram of the DSP Tx fast recovery method.

[0022] Fig. 9B DSP fast RX recovery manual mode flowchart, showing the normal operation of the module in DSP fast RX recovery manual mode.

[0023] Fig. 9C is a flow chart of the DSP Fast RX Recovery Manual Mode, showing the Initial Network Deployment - System Calibration Procedure. DETAILED DESCRIPTION

[0024] Again, the present disclosure relates to systems and methods for fast digital signal processor (DSP) optical receiver recovery, i.e., optical interconnects configured on the client side. The present disclosure is directed to DSP-based optical interconnects (note that terms such as transmitter / receiver, transceiver, transponder, pluggable optical module, plug, etc. are equivalent to optical interconnects). The present disclosure provides various methods for fast optical switching on the Tx and Rx sides of a link. Fast optical switching is suitable for shorter distance applications, such as applications using optical protection switches (OPS), transponder protection trays (TPT), 1+1, etc. The present disclosure relates to various types of modems, including non-return to zero (NRZ), pulse amplitude modulation (PAM), and coherent modulation (e.g., quadrature amplitude modulation (QAM), etc.). For example, the present disclosure can be implemented in QSFP-DD pluggable modules, QSFP-28 pluggable modules, etc.

[0025] DSP Adaptation

[0026] When the DSP-based Tx and Rx are switched from one fiber to another, or when switching occurs so that the Rx receives signals from different Tx, an adaptation process needs to be run, i.e., a control loop, a signal acquisition process, etc. In various types of DSP-based modems, i.e., NRZ, PAM, coherent, etc., the traditional adaptation process takes several seconds or even up to 10 seconds or more. For simple optical protection schemes, fast switching (millisecond level, e.g., 50ms or less) is required. Traditional methods will result in loss of signal (LOS) to higher layers (e.g., Ethernet) within a few seconds, and then the higher layers perform protection. This violates the purpose of simple optical protection schemes such as OPS, TPT, etc. Those skilled in the art will understand that the description of the present disclosure refers to OPS and TPT as optical protection schemes, but those skilled in the art will recognize that various other configurations are also considered (other applications may include the need to quickly recover from transient faults in the client Rx from other supporting equipment, for example, other protection schemes for routers or signal interruptions).

[0027] The adaptation process is performed to determine various settings on the Rx, including eye diagram optimization, continuous time linear equalization (CTLE), filter settings, etc.

[0028] The adaptation process can also be performed at the Tx to determine the settings there, such as transmit power, pre / post cursor settings, etc. The adaptation process must take into account the differences between the eyes of the two Txs. These differences may include amplitude, eye height, and distortion. For more complex modulation types, there are more losses that need to be equalized; for example, a PAM-4 signal requires more complex equalization than an NRZ signal. This complex approach requires sampling certain parameters over time before comparing to a threshold, which can be very time consuming.

[0029] Light protection

[0030] The application of fast optical switching is simple optical protection schemes, such as OPS, TPT, 1+1, etc. These schemes can be deployed in applications such as data centers, short-distance campus networks, and metropolitan area networks. The goal of these simple optical protection schemes is fast, efficient, and low-cost protection.

[0031] Figure 1 is a block diagram of the OPS module 10. Figure 21 is a block diagram of an OPS module 10 at a modem 12 on the Rx side. The OPS module 10 shows a transmit side 14 to two optical fibers C, D and a receive side 16 from two optical fibers E, F. On the Rx side, the OPS module 10 is able to perform switching based on optical power Rx, and only one of the optical fibers E, F will be connected to the optical fiber B through switching. On the Tx side, the optical fiber A from the Tx sends light to both optical fibers C, D. In one embodiment, the OPS module 10 performs switching based on optical power RX, for example, the transponder turns off its laser to initiate switching. In a working system, both lasers are on. During switching, port B will be lightless / unstable. Other switching methods are also contemplated, such as based on monitoring signals, looking for alarm indication signals (AIS), based on bit error rate (BER), etc.

[0032] The Rx side modem 12 includes an optical front end 18 and a DSP 20. There is a node 2 at a remote end from the Rx side modem 12. Using the OPS module 10, the node 2 includes two Tx modules 22, Tx1, Tx2. Figure 2 The time delay d between turning off the enabled Tx1 and activating the standby Tx2 or switching the switch 24 in the OPS module 10 is shown visually. (Note: it is not clear how it is illustrated because it appears that Tx2 is turning on).

[0033] Figure 3 is a block diagram of the TPT module 30 . Figure 4 is a block diagram of a TPT module 30 at the Rx side modem 12. The TPT module 30 is a 1+1 protection type, and its architecture is similar to the OPS module 10 at the Tx side and the coupler at the Rx side. For the OPS module 10, there is one Tx; for the TPT module 30, there are two Txs and on the Rx side, both lasers are never turned on at the same time.

[0034] These simple optical protection schemes are used to connect routers, switches, etc. or to connect subordinate network elements to switch clients. Again, the goal is to switch within 50ms and without sympathetic switching. Specifically, for both protection on the line and client protection on the associated client, there is a situation where client switching may cause anomalies to be detected in the line, for example due to a client module failure or a client network failure, causing undesired protection switching. This can be called sympathetic switching.

[0035] Optical switching requirements

[0036] For the applications described herein, up to 6.3 dB loss on a link supporting long reach (LR) applications or up to 18 dB loss on a link supporting extended reach (ER) applications is required.It is worth noting that the OPS or TPT modules 10, 30 suffer from a significant amount of loss.

[0037] These applications are able to tolerate short duration optical signal loss (LOS) / defects. This is a key requirement for the DSP20. During TPT or OPS switching in a back-to-back configuration, the Rx will experience short duration loss of light, clock loss, frame loss (or any unstable signal). The DSP / Rx needs to "overcome" these defects and not amplify them by entering a re-initialization phase (recovery typically must be <10ms from clearing the short duration signature). Currently, some plugs are found that amplify short duration signature LOS of a few milliseconds to a few seconds.

[0038] For OPS module 10, the laser (Tx) needs to be turned off quickly. Laser shutdown is used to initiate OPS switching. A mechanism needs to be provided to turn off the laser within 10ms (Integrated Circuit (I2C) register or SerDes fast Tx suppression mechanism). For TPT module 30, one laser needs to be turned off quickly and the other laser needs to be turned on quickly. TPT (Y-cable) switching is performed by turning off the laser on the currently active card and then turning off the laser on the newly active card. A mechanism also needs to be provided to turn off the laser within 10ms (Integrated Circuit (I2C) register or SerDes fast Tx suppression mechanism).

[0039] Quick recovery

[0040] The general approach to fast recovery is to avoid DSP 20 adaptation at the Rx and simply switch with the same settings or predetermined settings. In this way, protection switching to different fibers can be achieved within 50ms. That is, the DSP 20 just maintains the same settings or switches to predetermined settings. The basic assumption is that the application can support short-term LOS or defects without entering any restarts, optimizations, adaptations, control loops, etc. The traditional DSP approach sees a few milliseconds of light loss and amplifies it to many seconds to recover. After the fast switch, the DSP 20 at the Rx can make adjustments or changes, but the key here is that there is no automatic adaptation function, which may take several seconds.

[0041] In addition, another key requirement is that the DSP delays the loss of signal (LOS) or loss of light (LOL) reporting at Rx. This requirement enables the host device to avoid taking action on these faults before the adaptation delay is completed. This LOS / LOL reporting delay can be a configurable value.

[0042] Figure 5A is a TPT configuration diagram that illustrates the laser off / on and Rx recovery requirements. Figure 5B is based on Figure 5A Optical Rx characteristic diagram on the switch obtained with the configuration in . Figure 6 is a diagram showing three methods of DSP Rx fast recovery. Figure 7 is a TPT protection example diagram. The following examples describe fast recovery methods for TPT, but these methods are equally applicable to OPS and other protection switching methods. The key to this disclosure is that the functions at the Rx DSP and optionally at the Tx DSP, the protection switching and the network configuration can be anything.

[0043] Figure 5A , Figure 5B and Figure 6 An example network / test configuration is included with two Optical Transport Network (OTN) mappers with a service layer (Ethernet (Eth)) connected between them. On the right side, there is TPT protection from the client connected to the OTN mapper and to the adjacent router. Specifically, the OTN mapper includes two line interfaces connected to another OTN mapper, and two "plugs" which are client-facing interfaces connected to the router via TPT protection - two client Tx connected to one Rx. In the TPT configuration, both fast shutdown / opening of both client Tx and Rx recovery are required. In the OPS configuration, one client Tx will reach one Rx via two different paths and only Rx recovery is required. Of course, the example network / test configuration is given for illustrative purposes only, and one of ordinary skill in the art will recognize that various other configurations can be envisioned.

[0044] The present disclosure includes methods on both the Tx side and the Rx side. That is, fast Tx on / off is required for some configurations (e.g., TPT), but not necessarily for OPS, where OPS has no additional requirements on the Tx side. For both OPS and TPT, the DSP 20 has Rx recovery requirements.

[0045] Figure 9 shows a diagram of the Tx (off-ramp) direction recovery, which, like the Rx direction, must also be fast. In this case, the connection is electrical and signal source. The transceiver TX path must fully recover from TX unstable electrical signal events (clock transients, frame transients) at the host electrical input interface in <10ms.

[0046] Optical switching and fast recovery - transceiver requirements

[0047] Transceivers, modems, pluggable modules, etc. may have various requirements for optical switching and fast recovery. It should be understood by those skilled in the art that all of these requirements (labeled R1 - R5) are not necessarily required for the present disclosure.

[0048] R1. Link Budget - Supports links with up to 6 to 18 dB loss between plugs (channel insertion loss)

[0049] R2. Allow for short-term fiber LOS / defects (**Rx fast recovery is a key requirement for DSP vendors**) - During TPT or OPS switching, the Rx will experience short duration loss of light, clock loss, frame loss (or any unstable signal). The plug needs to "overcome" these defects and not amplify them by entering the reinitialization phase (recovery time must be <10ms from clearing the short duration signature). Currently, some plugs are found to amplify a few milliseconds of short duration signature LOS to many seconds.

[0050] R3. For OPS, fast laser shutdown is required - laser shutdown is used to initiate OPS switching - a mechanism must be provided to shut down the laser within 10ms (I2C register or SerDes fast Tx inhibit mechanism)

[0051] R4. For TPT, fast turning off and on of the laser is required - TPT (Y-cable) switching is performed by turning off the laser on the currently active card and then turning off the laser on the newly active card. A mechanism must be provided to turn the laser off or on within 10ms (I2C register or SerDes Fast Tx Squelch mechanism)

[0052] R5.TX Fast Recovery Time - The transceiver Tx path should recover from Tx unstable signal events (clock transients, frame transients) at the host electrical input interface in <10ms.

[0053] Laser off / on requirements and methods

[0054] For TPT, the Tx side is required to be able to close and open quickly. The fast closing and opening requirement is used when a Tx line is active but encounters an error that causes protection switching. As shown in Figure 5 and Figure 7 As shown, the active client Tx must be turned off at time t0, while the standby client Tx must be turned on. The requirement for fast Tx turn-off / on in TPT is because both Tx are connected to Rx. This is not the case in OPS, because only one Tx line is connected. Therefore, fast turn-off / on is not necessarily required in OPS configuration.

[0055] Here, fast can be on the order of 10 ms, ie, turned on or off in about 10 ms. There are several possible approaches.

[0056] 1) Use a fast Tx suppression mechanism, such as described in commonly assigned U.S. Patent No. 10,097,262, issued on October 9, 2018, entitled “Client Protection Switch in an Optical Pluggable Transceiver Activated by Fast Electrical Data Suppression,” the contents of which are incorporated herein by reference in their entirety.

[0057] 2) Use I2C (manage the communication to Tx to shut down). Some transmitters are Multi-Source Agreement (MSA) compliant and require this fast shutdown on the order of 10ms, while some MSA specifications may have more relaxed requirements. Again, see U.S. Patent No. 10,097,262 for a discussion of MSA requirements. The I2C approach needs to be superior to the MSA specification.

[0058] 3) Proprietary implementations where the DSP vendor, optical Tx vendor, etc. has configured the Tx to shut down for the desired time.

[0059] It is worth noting that this requirement for fast closing / opening is based on the fact that a large number of existing Tx configurations do not natively support this. Of course, there may be Txs that can natively support these specifications.

[0060] Rx switch signature and recovery requirements

[0061] The protection scheme is configured to switch between client Tx lines based on specific criteria. The Rx DSP 20 may not know which line it is receiving, but it does experience periods of no traffic, and the present disclosure defers any reacquisition at the RxDSP 20 during this period in the hope of enabling a fast switch, since any reacquisition may take several seconds. This Rx DSP functionality is independent of the underlying protection scheme.

[0062] The Rx recovery requirements are the same for OPS and TPT. The key here is to be able to withstand a period of poor traffic without causing it to take several seconds to reacquire the signal.

[0063] For OPS, protection switching is performed by the OPS module 10 or equivalent, where a 1x2 switch selects one of the two client Tx lines. The selection can be based on power, LOS, LOL, bit error rate, explicit commands, etc. During the switch, there is a period of time (t0 to t1 increment) where the Rx DSP 20 has no light / traffic.

[0064] For TPT, protection switching is actually performed at the client Tx, or based on the interruption of one of the client Tx lines, because the TPT module 30 or equivalent includes a coupler connecting the two client Tx lines to Rx.

[0065] For fast Rx recovery, the Rx DSP 20 is required to recognize the signature so that it knows that a protection switch is occurring, thereby letting the Rx DSP 20 know so that it can initiate a fast Rx recovery process, which essentially includes deferring any reacquisition and LOS / LOL reporting. Therefore, the Rx DSP 20 is configured to defer reacquisition and LOS / LOL reporting for a period of time. It is worth noting that this is a configurable aspect, that is, the Rx DSP 20 is in a fast mode configuration. For example, the Rx DSP 20 can be signaled using a vendor user space byte or through some other configuration.

[0066] The signature may include - Rx signature on switch (t0 to t1 increment), see Figure 5B .

[0067] 1) Signal conditions (this may depend on the client type, e.g. IEEE 802.3 Local Fault (Lf) if Ethernet, and other conditions and protocols considered)

[0068] 2) No light (up to 1 second, typical value = 15ms)

[0069] 3) There is light because protection switching has occurred

[0070] 4) Good clock and traffic – For asynchronous mapping, the clock can be skewed up to 9.2ppm from the original clock. For synchronous mapping, the clock will be the same. 400G will be synchronous. 100G can be synchronous or asynchronous. Ethernet allows clock variations up to ±100ppm, so 200ppm skew is possible between paths.

[0071] When the fast switching mode is enabled, the Rx DSP 20 is configured to detect the signature. Once detected, the Rx DSP 20 enters the fast switching mode and performs configuration settings and LOS / LOL delay until a predetermined period of time.

[0072] The Rx DSP 20 fast recovery should be completed within 10ms after receiving good traffic. For example, 400GigE is able to use FEC transparent monitoring mode for Rx LOL detection, such as using AM unlock for fast detection.

[0073] In one embodiment, the Rx recovery mode can be defined in the 100G, 400G, etc. module EEPROM vendor user space bytes. For example, the fast Rx recovery support advertising bit can include support for fast Rx automatic mode [on / off] and / or support for fast Rx manual mode [on / off]. The Rx recovery control bits can include 3 settings: standard compliant Rx recovery; fast Rx automatic mode; fast Rx manual mode. The fast Rx manual mode setting bits can include Tx1 / Tx2 source selection (default Tx1); Rx DSP settings for Tx1; and Rx DSP settings for Tx2. Further control bits include (DSP adaptation delay (0 to 1 second), LOS report delay (0 to 1 second).

[0074] Fast Rx recovery mode should be within about 10ms after good traffic (stable signal). Fast RX Auto mode includes DSP recovery methods 1 and 3. Fast Rx Manual mode includes DSP recovery method 2.

[0075] Method 1: Default starting point (sufficient to carry traffic but not necessarily optimal). When transients occur, use defaults and allow optimization / training to occur later. Defaults can be vendor standards or factory calibration for all plugs.

[0076] 1) The DSP chip saves a factory default starting point when calibrated. It is sufficient to carry traffic but not necessarily optimal. When transients occur, use the default value and allow for optimization / training later.

[0077] 2) When the traffic interruption (fault detection) lasts longer than the specified adaptation delay, the DSP should perform a complete retrain of the current link. When shorter than the adaptation delay, the DSP will obtain the saved factory default parameters and restore these values. This option may require stricter engineering rules than option 2 in terms of matching uplink Rx power, ppm offset, temperature, etc. Equipment may need to be co-located with this option. Traffic interruption (fault detection): fast detection is required.

[0078] Method 2: Ability to learn (train) on each path to determine any DSP recovery value (stored Tx1 and Tx2 values).

[0079] 1) The DSP chip saves the parameters after the CDR converges during the training phase. It saves two sets of parameters - one for Tx1 and the other for Tx2.

[0080] 2) When the traffic interruption (fault detection) lasts longer than the specified adaptation delay, the DSP should perform a complete retrain of the current link. When the traffic interruption lasts shorter than the specified adaptation delay, the DSP does not retrain. Instead, the DSP fetches the saved parameters for the new link and applies them to the recovery (Tx1 if Tx2 was used last, and Tx2 if Tx1 was used last).

[0081] 3) Although the DSP tunes parameters to compensate for voltage / temperature changes when the link is up, the parameters saved are only the result of the initial training, because when the link is restored, the voltage / temperature is unknown anyway.

[0082] Traffic interruption (fault detection): requires fast detection.

[0083] NOTE: The Fast RX Recovery method will require a manual Fast Module Rx Recovery calibration procedure to determine and store the optimal Rx DSP settings required for both Tx1 and Tx2. This calibration process will be performed at the system level. To complement the system level calibration, there is also a module level post calibration procedure in manual mode. See Fig. 9B and 9C This post calibration allows for automatic TX1 / TX2 calibration if a system level calibration did not occur or did not occur correctly for some reason.

[0084] The proposed manual fast RX recovery system calibration method ( Fig. 9C ).

[0085] 1. Host triggers manual fast module Rx recovery calibration

[0086] 2. Module for verifying support for and switching to Fast Rx Manual mode

[0087] 3. Calibrate the receiver using Tx1 (frame traffic) and store the Rx DSP settings in the Rx DSP settings allocated for the Tx1 module EEPROM bytes

[0088] 4. Calibrate the receiver using Tx2 (frame traffic) and store the Rx DSP settings in the Rx DSP settings allocated for the Tx2 module EEPROM bytes

[0089] 5. Set Tx source to Tx1; and apply Rx DSP settings to Tx1

[0090] 6. Terminate the calibration procedure.

[0091] Recommended manual fast RX recovery module post-calibration method ( Fig. 9B )

[0092] 1. DSP detects DSP RX LOS

[0093] 2. If DSP LOS <= Adaptation Delay, assume this event is a protection switch and apply alternate Tx settings (if on TX1 then apply TX2, if on TX2 then apply TX1)

[0094] 3. If DSP LOS > DSP Adaptation Delay, then assume this is not a protection switch (same Tx still connected). In this case, apply the complete reacquired Tx settings to the current TX (to TX1 if on TX1, to TX2 if on TX2).

[0095] Method 3: Hold mechanism: Use the value from the Tx you switched away from (eg, if switching to Tx2, use Tx1).

[0096] 1) When a traffic interruption (fault detection) lasts longer than the specified adaptation delay, the DSP will do a full adaptation recovery (involving retraining). When a traffic interruption lasts shorter than the specified adaptation delay, the DSP will not retrain. Instead, the DSP simply uses the current values ​​for recovery. For example, if switching from Tx1 to Tx2, the current values ​​of Tx1 will be used.

[0097] 2) This option may require tighter engineering rules than option 2 in terms of matching uplink Rx power, ppm offset, temperature, etc. Equipment may need to be co-located with this option.

[0098] The above DSP recovery methods 1 and 2 require fast fault detection.

[0099] Configurable adaptation delay can be from 0 to 1 second in 100ms steps, or other values. This can include configurable insertion delay for LOS / LOL reporting (to prevent the host device from taking action on these faults before the adaptation delay has completed)

[0100] Since the standard plug interface for signaling plug LOS is slow (via I2C), this signal is not fast enough for DSP fault detection under methods 1 and 2, where DSP fault detection triggers switching to use the alternate recovery value. Method 3 (fast RX recovery hold mechanism) does not require fast fault detection.

[0101] Fast fault detection

[0102] Possible fast fault detection techniques are as follows (fast fault detection is required for fast Rx recovery methods 1 and 2)

[0103] Fast Fault Detection Technique 1: Use FEC / PCS Blocks in DSP for Ethernet Rates - Use the Physical Coding Sublayer (PCS) / Forward Error Correction (FEC) blocks in DSP for Ethernet rates (example below):

[0104] 400GE: Ability to use FEC transparent monitoring mode for Rx LOL detection, such as AM unlocking for fast detection. In this case, FEC / PCS is not terminated (retimer mode), but can be monitored for failure. PCS layer failure detection is very fast (a few milliseconds), so it can be used as a trigger for fast recovery when the DSP needs to switch to a different recovery value.

[0105] 100GE: Ability to use FEC termination mode for Rx LOL detection. For example, AM unlocked for fast detection. In this case, FEC / PCS is terminated and can be monitored for faults. PCS layer fault detection is very fast (a few milliseconds), so it can be used as a trigger for fast recovery when the DSP needs to switch to a different recovery value.

[0106] Fast Fault Detection Technique 2: Use Available Plug Reserve Pins - For interfaces without FEC or Framing Blocks, use the plug reserve pins as follows:

[0107] The host card detects the failure (via the framer) and drives the reserved pin to quickly notify the DSP. This will solve the problem of plug LOS / LOL detection via I2C being too slow (up to 100ms). OTU4 does not have an FEC transparent monitoring option

[0108] For QSFP-DD, the following pinouts can be used

[0109] Pin 66 is used for VREF_TEST: MSA RESERVED pin.

[0110] QSFP-DD emulates VREF_TEST pin for factory test host. Includes series 10k resistor. On normal host, leave this pin floating or connect 10k to GND. Another reserved pin in CMIS5.1 (pin 46) is now Vs (vendor specific) pin.

[0111] Other possible fault signaling methods include modulating a code word on a low-speed signal such as ePPS, modSel, or vendor-specific pins VS1 to Vs4.

[0112] Fast fault detection technique 3: Photocurrent signal loss detection - rate agnostic.

[0113] Configuration settings

[0114] Once the Rx DSP 20 detects the fast switching pattern, the Rx must maintain or change its configuration settings. Here, the Rx DSP 20 first acquires light from the working Tx client link, stops acquiring light and detects the fast switching pattern, and then starts acquiring light from the protection Tx client link. The Rx DSP 20 may have to take some actions to change the settings of its working Tx client link and the protection Tx client link. Note that the terms "working Tx client link" and "protection Tx client link" are used for illustration. These can be any type of fiber links.

[0115] There are several options. The following list is some example options, other options are also envisioned. Optical modules (such as pluggable) will have 2 fast recovery option modes, such as Figure 5A and 5B As shown (Fast Rx Auto Mode and Fast Rx Manual Mode). In Manual Mode, it can store TX1 and TX2 settings.

[0116] 1) Fast Rx Auto Mode. Change to default settings or continue to operate with default settings, i.e., sufficient to carry traffic but not necessarily optimal. When transients occur, use default settings and then allow optimization / training afterwards. The default setting values ​​can be vendor standards, factory calibration, configuration, etc. for all plugs. In one embodiment, the default settings are used for both Tx client links. In another embodiment, the working Tx client link is optimized to the optimal settings and the Rx DSP 20 switches to the default settings based on the fast switching mode.

[0117] 2) Fast Rx Auto Mode Hold, where the optimal settings of the working Tx Client link are used on the new protection Tx Client link. For example, if switching to the protection Tx Client link, the optimal settings of the working Tx Client link are used.

[0118] 3) Fast Rx Manual Mode. Change some predefined settings of the protection Tx client link. Here, for example, at configuration time, the settings of both the working Tx client link and the protection Tx client link may be determined, and the settings of the protection Tx client link are stored and used when entering the fast switching mode.

[0119] There can also be a configurable adaptation delay setting, for example, from 0 to 1 second in 100ms steps. After the adaptation delay setting and bad traffic expires, the Rx DSP 20 can start signal acquisition. Bad traffic may be based on certain BER thresholds or links. In addition, after the adaptation delay and bad traffic expires, the Rx DSP 20 can report LOS / LOL so that any host device can start taking action on these faults.

[0120] Example Implementation

[0121] 1) When traffic on the current link is interrupted (eg, by LOS), the optical component notifies the Rx DSP 20.

[0122] 2) When traffic on the other link is restored, the optical component notifies the Rx DSP 20.

[0123] 3) The Rx DSP 20 saves the parameters after the convergence of the clock and data recovery (CDR) during the training phase. It saves two sets of parameters - one for TX1 (working Tx client link) and the other for TX2 (protection Tx client link).

[0124] 4) When the traffic interruption lasts for more than 1 second, the Rx DSP 20 should perform a complete retrain of the current link. Please note that the term "retrain" is the same as adaptation, optimization, etc. When the traffic interruption lasts for less than 1 second, the Rx DSP does not retrain. Instead, the Rx DSP 20 obtains the saved parameters of the new link and applies (if TX2 was used last, TX1 is applied, and if TX1 was used last, TX2 is applied). This is to achieve a relink time of 10ms.

[0125] 5) Although the Rx DSP 20 tunes parameters to compensate for voltage / temperature changes when the link is up, the parameters saved are only the result of the initial training, because when the link is restored, the voltage / temperature is unknown anyway.

[0126] Traffic interruption detection for fast recovery may need to come from the Rx DSP 20 rather than relying on the plug optical LOS, as this may be too slow to detect.

[0127] Another way to improve Rx recovery time is to use two transmitters that are configured to operate with the same Rx settings, either through calibration, configuration-time adjustments, or periodic adjustments of Tx parameters. One approach is to select two transmitters that provide similar eyes so that no adjustments are required. For example, this could include similar transmitters, such as identical modules from the same vendor. Alternatively, the transmitters can be configured to operate with common Rx settings.

[0128] Rx DSP Process

[0129] Figure 8is a flow chart of a process 100 that may be implemented by a circuit such as Rx DSP 20. Process 100 includes detecting a traffic interruption on a current link associated with an optical receiver (step 102), entering a delay period (step 104), and performing any of the following operations: i) receiving good traffic during the delay period; ii) expiring the delay period to notify a host device and retraining to acquire an optical signal (step 106).

[0130] Process 100 can also include: in response to the delay period, avoiding retraining the optical signal and avoiding any notification to the host device (step 108). Process 100 can also include, in response to the delay period, changing the settings of the optical signal to one of a default setting and a predetermined setting for the new link (step 110). The settings of the optical signal can be maintained during the delay period. The notification can include loss of signal (LOS) and loss of light (LOL).

[0131] Process 100 can also include receiving a configuration for a fast switching mode using a delay period.Process 100 can also include, in response to good traffic within the delay period, continuing to process the good traffic and retraining the optical signal while processing the good traffic (step 112).

[0132] The optical receiver can be connected to an optical protection switch (OPS). The optical receiver can be connected to an optical coupler so that two different links including the current link can send optical signals; this can be a TPT. The corresponding optical transmitters on the two different links can be configured for fast switching on and off. The optical signal can include non-return-to-zero (NRZ) modulation, pulse amplitude modulation (PAM), or coherent modulation.

[0133] TX fast recovery time requirement R5 detailed description

[0134] exist Fig. 9A In the , the Tx fast recovery requirement means that the transceiver Tx path should recover from Tx instability events at the host electrical input interface (clock transients, frame transients) in <10ms. Vendor Optional (if required): Similar to the Rx fast recovery mode described previously, a fast Tx recovery mode can be implemented if needed and appropriate.

[0135] Tx signature at recovery (increment from t0 to t1)

[0136] 1. Signal conditions (depending on the client type. For example, if it is Ethernet, it is 802.3LF)

[0137] 2. Data rate ppm deviation (transient)

[0138] 3. Electrical inhibition or invalid signal from the host (maximum 1 second, typical 8ms to 15ms)

[0139] 4. Asynchronously mapped clock and traffic are good (clock may be offset by 200ppm from original clock)

[0140] For synchronous mapping, the clocks will be the same

[0141] 400G or 100G can be synchronous or asynchronous, depending on the mapping used

[0142] in conclusion

[0143] It is worth noting that some embodiments described herein may include or use one or more general or special processors ("one or more processors"), such as a microprocessor; a central processing unit (CPU); a digital signal processor (DSP); a custom processor, such as a network processor (NP) or a network processing unit (NPU), a graphics processing unit (GPU) or the like; a field-programmable gate array (FPGA); and unique stored program instructions (including software and firmware) for controlling it, combined with certain non-processor circuits to implement some, most or all of the functions of the method and / or system described herein. Alternatively, some or all of the functions may be implemented by a state machine without stored program instructions, or in one or more application-specific integrated circuits (ASICs), where each function or certain combinations of certain functions are implemented as custom logic or circuits. Of course, a combination of the above methods may also be used. For some embodiments described herein, the corresponding devices in hardware and optionally with software, firmware and combinations thereof may be referred to as "circuits configured as...", "logic configured as...", etc., to perform a set of operations, steps, methods, processes, algorithms, functions, techniques, etc. described herein for various embodiments on digital and / or analog signals.

[0144] In addition, some embodiments may include a non-transitory computer-readable medium having stored thereon instructions for programming a computer, server, device, apparatus, at least one processor, circuit / circuit group, etc. to perform the functions described and claimed herein. Examples of such non-transitory computer-readable media include, but are not limited to, hard disks, optical storage devices, magnetic storage devices, read-only memories (ROMs), programmable ROMs (PROMs), erasable PROMs (EPROMs), electrical EPROMs (EEPROMs), flash memory, etc. When stored in a non-transitory computer-readable medium, software may include instructions executable by one or more processors (e.g., any type of programmable circuit or logic), which, in response to such execution, cause the one or more processors to perform a set of operations, steps, methods, processes, algorithms, functions, techniques, etc. described herein for various embodiments.

[0145] Although the present disclosure has been illustrated and described herein with reference to preferred embodiments and specific examples thereof, it will be apparent to one of ordinary skill in the art that other embodiments and examples may perform similar functions and / or achieve similar results. All such equivalent embodiments and examples are within the spirit and scope of the present disclosure, are contemplated thereby, and are intended to be covered by the following claims. In addition, it is noteworthy that the various elements, operations, steps, methods, processes, algorithms, functions, techniques, etc. described herein may be used in any combination with one another.

Claims

1. A receiver Rx digital signal processor DSP (20) in an optical receiver, the Rx DSP (20) comprising a circuit, wherein the circuit is configured to: detecting (102) a traffic interruption on a current link associated with the optical receiver, Entering (104) a delay period in which no loss of signal LOS or loss of light LOL is reported while the optical signal associated with the traffic i) switches to a new link or ii) recovers from an outage on the current link, and Perform one of the following (106): i) receiving traffic within the delay period; or ii) expiring the delay period, thereby notifying a host device and retraining the Rx DSP (20) to acquire the optical signal to receive the traffic on a new link or a current link, wherein the Rx DSP (20) is retrained to determine various settings on the optical receiver.

2. The Rx DSP (20) according to claim 1, wherein the circuit is configured to During the delay period, retraining to acquire the optical signal is avoided (108) and any notification to the host device is avoided.

3. The Rx DSP (20) according to any one of claims 1 to 2, wherein the circuit is configured as After entering the delay period, the settings for the optical signal are changed (110) to one of a default setting or a predetermined setting for the new link.

4. The Rx DSP (20) according to any one of claims 1 to 2, wherein: The settings for the optical signal are maintained during the delay period.

5. The Rx DSP (20) according to claim 1 or 2, wherein: The notification includes one or more of loss of signal, LOS, and loss of light, LOL.

6. The Rx DSP (20) according to claim 1 or 2, wherein: The circuit is also configured to receive a configuration in which the delay period is used.

7. The Rx DSP (20) according to claim 1 or 2, wherein: The circuit is also configured to, in response to (112) receiving the traffic within the delay period, continue processing the traffic and retrain the optical signal while processing the traffic.

8. The Rx DSP (20) according to claim 1 or 2, wherein the optical receiver is connected to an optical protection switch (OPS).

9. The Rx DSP (20) according to claim 1 or 2, wherein the optical receiver is connected to an optical coupler, whereby two different links including the current link can transmit the optical signal.

10. The Rx DSP (20) of claim 9, wherein the corresponding optical transmitters on the two different links are configured to be quickly turned off and on.

11. The Rx DSP (20) according to claim 1 or 2, wherein the optical signal comprises non-return-to-zero (NRZ) modulation.

12. The Rx DSP (20) according to claim 1 or 2, wherein the optical signal comprises a pulse amplitude modulation (PAM).

13. The Rx DSP (20) according to claim 1 or 2, wherein the optical signal comprises coherent modulation.

14. The Rx DSP (20) according to claim 1 or 2, wherein: The transmitter for the current link and the transmitter for the protection link setup are adjusted so that the Rx DSP does not need to be adjusted when switching between links.

15. A method (100) implemented in a receiver Rx digital signal processor DSP (20) in an optical receiver, the method comprising the following steps: detecting a traffic interruption (102) on a current link associated with the optical receiver, entering a delay period (104) in which no loss of signal (LOS) or loss of light (LOL) is reported while the optical signal associated with the traffic i) switches to a new link or ii) recovers from an outage on the current link, and Perform one of the following (106): i) receiving traffic within the delay period; or ii) expiring the delay period, thereby notifying a host device and retraining the Rx DSP (20) to acquire an optical signal to receive the traffic on a new link or a current link, wherein the retraining includes determining various settings on the optical receiver.

Images