The present application claims priority from us patent application 61/668,690 filed on 7/6/2012 and us patent application 13/906,956 filed on 5/31/2013, which are incorporated herein by reference in their entirety.
Detailed Description
Various embodiments of the invention are discussed in detail below. While specific implementations are discussed, it should be understood that these implementations are for illustrative purposes only. A person skilled in the relevant art will recognize that other components and configurations may be used without parting from the spirit and scope of the invention.
Power-saving ethernet networks attempt to save power when the network's traffic utilization is not at its maximum capacity. This serves to minimize performance impact while maximizing energy savings. One type of traffic profile that may be encountered by the energy saving control strategy is an asymmetric traffic profile. In one example, asymmetric traffic profiling may be found in an automotive system environment, where a first link direction carries video traffic and a second link direction carries infrequently used low bandwidth status, protocol, diagnostic, and/or control traffic.
Asymmetric traffic profiling may limit the amount of savings that can be obtained. For example, the presence of traffic in either direction of the presence link may prevent the power saving control strategy from causing the link to enter a low power mode, such as a low power idle mode. According to the present invention, an energy saving control protocol is provided such that the energy saving can be increased when an asymmetric traffic profile is encountered.
In one embodiment, the power saving method of the present invention may be configured to control the operation of a physical layer device (PHY) in a network device. In one example, a PHY in a network device is initially configured to operate in a full duplex mode, where the PHY transmits and receives data over a transmission medium. In various applications, the transmission medium may represent twisted wire pairs, fiber optic cables, backplanes, and the like.
When operating in full duplex mode, it may be determined by the power saving control strategy that a transmission in the first direction has entered a low link utilization condition in full duplex mode. Based on such a determination, the power-saving control policy may be configured to transition the PHY from a full-duplex mode to a simplex mode in which transmission in a first direction over the transmission medium is interrupted while transmission in a second direction over the transmission medium continues. Power savings can be achieved in simplex mode. For example, the circuitry used during full duplex mode for echo cancellation, near end crosstalk (NEXT), far end crosstalk (FEXT), alien near end crosstalk (ANEXT), alien far end crosstalk (AFEXT), Transmit (TX) DSP, Receive (RX) DSP, pre-emphasis, etc., may be disabled, the power to the segments disconnected, or reduced usage during PHY operation in simplex mode.
In one embodiment, it may be further determined by the energy saving control strategy that the transmission in the second direction used in simplex mode has also entered a low link utilization condition. Based on such further determination, the power-saving control policy may be further configured to transition the PHY from a simplex mode to a low-power mode in which both the first direction of transmission and the second direction of transmission over the transmission medium are interrupted. Such further transitions may be embodied by a low power mode, such as a Low Power Idle (LPI) mode. In general, LPI depends on opening an active channel that is inactive in both transmission directions when there is no data transmission. Thus saving energy when the link is down. A refresh signal may be periodically sent to enable wake-up from LPI mode.
The invention is characterized in that the power saving control protocol can be designed to control the transition of states between full duplex mode, simplex mode and low power mode. Here, the full-duplex mode refers to a transmission mode in which two PHYs are transmitted on a transmission medium (e.g., twisted pair, fiber optic cable, backplane, etc.), the simplex mode refers to a transmission state in which only one PHY is transmitted on the transmission medium, and the low-power mode refers to a transmission state in which any PHY is transmitted on top of the transmission medium. In operation, a full-duplex mode may represent a normal operating state, a simplex mode may represent an operating state in which only one of the two PHYs is ready to enter a low power mode, and a low power mode may represent a low power mode in which both PHYs are ready to enter a low power mode. It is a feature of the present invention that further energy savings can be achieved by disabling the PHY cancellation circuitry used in full duplex mode, but not required in simplex mode.
Before describing the details of an energy saving control protocol using the present invention, a description of an energy saving control strategy that can be used to implement the present invention is first provided. Broadly, a power saving control policy for a particular link in a network determines when to enter a power saving state, what power saving state (e.g., power saving class) to enter, how long to remain in a power saving state, to which power saving state to transition from a previous power saving state, whether to affect traffic profiling (e.g., merge packets, buffer and batch processing, rebalance traffic, shape traffic, etc.), and the like. In one embodiment, the energy saving control policy may base these energy saving decisions on a combination of settings established by the IT administrator and traffic characteristics on the link.
Fig. 1 shows an exemplary link to which the power saving control strategy of the present invention may be applied. As shown, the link supports communication between the first link partner 110 and the second link partner 120. In various embodiments, the link partners 110 and 120 may represent switches, routers, endpoints (e.g., servers, clients, VOIP phones, wireless access points, etc.), and the like. As shown, the link partner 110 includes a PHY112, a Media Access Control (MAC) 114, and a host 116, while the link partner 120 includes a PHY122, a MAC124, and a host 126.
In general, the hosts 116 and 126 may comprise suitable logic, circuitry, and/or code that may enable operability and/or functionality of the five highest functional layers of data packets to be transmitted over a link. Since each layer in the OSI model provides services to the directly higher interface layer, the MACs 114 and 124 can provide the necessary services to the hosts 116 and 126 to ensure that the packets are properly formatted and transmitted to the PHYs 112 and 122, respectively. The MACs 114 and 124 may comprise suitable logic, circuitry, and/or code that may enable handling of data link layer (layer 2) operability and/or functionality. The MACs 114 and 124 may be configured to implement ethernet protocols, such as those based on the IEEE802.3 standard. PHYs 112 and 122 may be configured to handle physical layer requirements including, but not limited to, packetization, data transmission, and serialization/deserialization (SERDES).
As further shown in fig. 1, the link partners 110 and 120 also include energy saving control policy entities 118 and 128, respectively, that are configured to implement energy saving control protocols that transition between full duplex mode, simplex mode, and low power mode. In general, the energy saving control policy entities 118 and 128 may comprise suitable logic, circuitry and/or code that may enable establishing and/or enforcing energy saving control policies of network devices. In various embodiments, the energy saving control policy entities 118 and 128 may be logic and/or functional modules that may be implemented in one or more layers, including portions of a PHY or other subsystems in an enhanced PHY, MAC, switch, controller, or host, to enable energy saving control at one or more layers.
Conventional power saving control protocols may be designed to determine that a low power mode may be entered when there is no data transmission in both directions of the link. An example of such a low power mode is the LPI mode, in which both transmitters are not active except for a short period of the refresh signal. The use of LPI mode is in contrast to conventional idle signaling when no data needs to be sent. It will be appreciated that the transmission of a conventional idle signal will consume as much power as possible for data transmission. Another example of a low power mode is a subset PHY mode, where one or more channels of the PHY device are reconfigurable in real-time or in real-time to communicate at different data rates.
For link applications such as gigabit ethernet (1000 BASE-T), traffic present on either end of the link will prevent the link from entering a low power mode such as LPI mode. Here, one side of the link will transmit data, while the other side of the link will transmit idle signals. This situation is shown in fig. 2. As this case illustrates, the presence of stable data at one end of the link will prevent the other end of the link from entering a low power idle mode.
The inefficiency of this situation is inherent in the two-way protocol where data present on either end of the link will prevent the link itself from entering the low power idle mode. As shown in fig. 3, data transmitted from either end of the link generally does not occur simultaneously. Because there is no correlation between data arriving at one end of the link and data arriving at the other end of the link, the idle state of the link is based on an and function of the idle availability of transmissions in both directions on the link. Thus, even if one end of the link is nearly 100% idle, traffic appearing at the other end of the link will preclude power savings through low power modes. In the 1000BASE-T current specification, for example, idle signals are sent during the time when the other end of the link is sending data. Thus, entering LPI mode depends on no data being transmitted in both directions of the link. Such an operation represents an energy saving protocol such as 1000 BASE-tee.
In the present invention, it is recognized that asymmetric traffic profiling may severely limit energy saving opportunities. This is especially true for automotive networks where most links involving sensors, controllers, entertainment systems, etc. operate in a very asymmetric manner. Another example of a network with asymmetric traffic profiles includes an audio-video bridging (AVB) network that transmits streaming media traffic (e.g., to access the network to home/business, cellular backhaul communications, etc.), and a control network, where one direction of the link is control information with a light traffic profile and another direction of the link is status information with a heavy traffic profile. Here, it should be noted that these conditions are dynamic to the user, but appear relatively static in terms of network timing. For example, with a home computer or device a user may start a movie stream that lasts for hours and then return to a normal workflow where the traffic on the link is profiled completely differently. It is therefore a feature of the present invention that additional simplex operation modes, full duplex modes and low power modes may be defined for use by the energy saving control strategy.
Here, it should be noted that the simplex mode may be used in the context of any transmission system that uses full duplex mode communication. Consider, for example, data transmission over a single twisted pair (e.g., 100Mbps, 1Gbps, or other standard or non-standard speeds). In this example, a single twisted pair would be used to transmit and receive data in full duplex mode. On the other hand, in simplex mode, a single twisted pair would be used to transmit data in only one direction. Figure 4 illustrates the operation of this simplex mode. When operating in such simplex mode, the echo canceller may be switched off because transmission is only in one direction over a single twisted pair.
Fig. 5 illustrates an exemplary state diagram used by the power saving control protocol of the present invention to handle links with asymmetric traffic profiles. As shown, the state diagram includes a full duplex mode 510, a simplex mode 520, and a low power mode 530, and corresponding transitions to/from the operational modes 510, 520, 530. Full-duplex mode 510 may represent a normal operating state in which both PHYs on the link transmit over a transmission medium (e.g., twisted pair, fiber optic cable, backplane, etc.). The transition from the full duplex mode 510 to the low power mode 530 may be made based on a transition 'a', where the two link partners declare a transition to the low power mode 530. As an example, the low power mode may represent a conventional low power mode, such as an LPI mode or a subset PHY mode. It will be appreciated that the low power mode may represent a different form of low consumption mode on the link that can save power. It should again be noted that the transition from full duplex mode to low power mode is limited to the case where no data is transmitted at either end at a common point in time. As shown in fig. 3, the presence of data transmitted by any link partner will preclude entry into the low power mode. In one embodiment, a simplex LPI or simplex subset PHY combined mode may also be included between the simplex mode and the low power mode.
When there is an asymmetric traffic profile, only one link partner may declare a transition to the low power mode while the other link partner continues regular traffic transmission on the transmission medium. This situation may be represented by a transition 'B' from duplex mode 510 to simplex mode 520. In a state transition from full duplex mode 510 to simplex mode 520, only one PHY on the link transmits on the transmission medium. For example, in a single twisted pair implementation, only a single PHY transmits on a single twisted pair. As described above, simplex mode 520 may save energy because one or more of the echo cancellation, NEXT, FEXT, ANEXT, AFEXT, TX DSP, RX DSP, pre-emphasis, etc. circuits used in full-duplex mode 510 may be disabled, the power to the segments disconnected, or the use reduced.
The same link partner that announces the transition to the low power mode cancels the transition and a transition from the simplex mode 520 back to the full duplex mode 510 will then occur. This transition is represented by transition 'C' in the state diagram of fig. 5, and the link will return to a normal operating state.
On the other hand, if the link is operating in simplex mode 520 and the other link partner declares a transition to low power mode, then a situation occurs where both link partners have declared such a transition. This represents a situation where neither link partner has data to transmit. Once this occurs, the power saving control protocol will initiate a transition from the simplex mode 520 to the low power mode 530. This conversion is shown as conversion 'D'.
In one embodiment, the transition between the full duplex mode 510 to the simplex mode 520 may depend on a waiting period. The wait period may be designed to detect whether a transmission in the other direction is signaling a transition to the low power mode, thereby causing a direct transition from the power duplex mode 510 to the low power mode 510. This waiting period may remove unnecessary transitions of the simplex mode 520 as intermediate transitions to the low power mode 530. In one embodiment, applied to non-delay sensitive applications, the control policy may choose to buffer traffic to keep it in a particular mode for an extended period of time (e.g., if transitioning to a higher power mode), or to ensure that changes in the profile are not permanent (e.g., a temporary idle state where the cost of switching modes would be higher than the energy saving benefit).
While in the low power mode 530, if any of the link partners cancels the transition to the low power mode, a transition from the low power mode 530 to the simplex mode 520 may occur. This conversion is represented by conversion 'E', where only one PHY transmits on a conductor. Alternatively, if both link partners cancel transitioning to the low power mode at times close to each other, a transition from the low power mode 530 to the full duplex mode 510 may occur. This transition is represented by transition 'F' where normal operation will resume on the link.
As described above, the simplex mode introduced in addition to the full-duplex mode and the low-power mode may result in a significant amount of power savings through the power saving control protocol when the link exhibits asymmetric traffic profiling while the conventional low-power mode is not functional.
Fig. 6 shows another exemplary application for data transmission over two twisted pairs. In this example, two twisted pairs are used in full duplex mode, where each twisted pair may be used to transmit and receive 500Mbit/s of data. In simplex mode, for example, once link partner B declares a transition to low power mode or disables or reduces one direction of the link, PHY B will no longer transmit on twisted pair a or B. Further, PHY A may be configured to transmit 1Gbit/s of data over twisted pair A as compared to 500Mbit/s over twisted pairs A and B. Crosstalk cancellers (e.g., NEXT and FEXT) may also be disabled because transmission occurs in only a single direction on one twisted pair, thereby saving additional power.
Although the above example shows twisted pair B not being used, twisted pair B may also be used to transmit refresh signals. In addition, twisted pair B can also be used to transmit from PHY B to PHY a at a low transmission rate, accommodating low power, low loss modes. Here, the low-loss mode enables a data transmission rate substantially less than the nominal data transmission rate of 500Mbit/s over twisted pair B.
The principles of the present invention may be applied to different PHY types (e.g., twisted pair, fiber optic cable, backplane, etc.), different interface types such as standard wiring or non-standard wiring (e.g., automotive wiring and other control networks), shared media and associated interfaces, e.g., EPON, xPON, EpoC, ethernet over DSL, etc. Furthermore, the principles of the present invention may be applied to different standardized and non-standardized data transmission speeds.
Another embodiment of the invention may provide a machine and/or computer readable memory and/or medium having stored thereon a machine code and/or computer program having at least one code portion executable by a machine and/or computer to cause the machine and/or computer to perform the steps described herein.
These and other aspects of the invention will be apparent to those skilled in the art from a review of the foregoing detailed description. While a number of the salient features of the invention have been described above, it will be apparent to those of ordinary skill in the art, after reading the present disclosure, that the embodiments of the invention may be implemented and carried out in a variety of ways, and thus, the above description should not be taken as excluding such other embodiments. Also, it is to be understood that the phraseology and terminology employed herein is for the purpose of description and should not be regarded as limiting.