Patent Application
20230216538
This present invention relates to high-speed synchronous data transmission systems that use multiple signal subcarriers, such as those operating over Digital Subscriber Lines (DSL). More particularly, this invention is directed to the balance of global and local performance optimizations of DSL systems, especially Very High-Bit-Rate Digital Subscriber Line (VDSL) and future variants that suffer more from crosstalk in the whole system.
Digital Subscriber Lines (DSL) have been popular since ADSL (Asymmetric DSL) was invented and standardized in 1999. It was a great technology leap from a voice band modem, which only utilizes the voice band up to 4 KHz. In many countries in the world, it gained immediate popularity to provide broadband internet to each home, with the fact of wide existing deployment of telephone wires. The broadband speed was lifted from a mere 50 Kbps by a voice band modem, to 8 Mbps by an ADSL modem with 1 MHz bandwidth. Throughout the years, DSL technology continued to advance. With more advanced technology, more and higher bandwidths are used to increase the attainable speeds. Baseband bandwidth started from 1 MHz to 2 MHz (ADSL2+), to 8 MHz/17 MHz/35 MHz (VDSL2), and to 106 MHz/212 MHz (GFAST). The bandwidth will be evenly used by a set of subcarriers with orthogonal frequencies, and such technology is called Discrete Multi-Tone (DMT). Along with the technology advances, it was found that the crosstalk among telephone wires in the same cable bundle posed an increasing problem when the using bandwidth got higher. To achieve an overall optimal system performance, a later technology called vectoring was invented to cancel most of the crosstalk within the same DSL technology. Crosstalk can be evaluated on subcarrier frequency level within the same bandwidth. With a good estimation of crosstalk, the Digital Subscriber Line Access Multiplexer (DSLAM, residing in a central office) side may do a good cancellation on most of the unwanted crosstalk. This topic is becoming more and more important as the involving bandwidth will only go up higher for uprising new technologies, and the crosstalk will get more and more serious.
In order to deal with this increasing crosstalk interferences among the whole DSL system, several ideas have been developed and implemented. It is called the Dynamic Spectrum Management (DSM), which was mainly contributed by Stanford professor John Cioffi and his team. The techniques for DSM are categorized into levels of coordination. In Level 0, there is no coordination, and each user views other users' signals as noise and seeks to maximize its own data rate in a distributed manner. This is called Iterative Water-filling (IWF). The next is Level 1, a Spectrum Management Center (SMC) at DSLAM side can coordinate some power back-off for short-distance users located close to central office (CO) as it is not needed to reach their service rates. This in turns reduces the crosstalk into other longer-distance users, who need the full power for their service rates. Then it goes to Level 2, when the SMC can coordinate the spectra of all modems centrally. It applies so-called Optimal Spectrum Balancing (OSB), attempting to maximize the weighted sum of rates of all users. SMC may decide both upstream and downstream PSDs to achieve that. For Level 3 DSM, there is the complete coordination, or ‘vectoring’ occurs as all modems terminate at the same DSLAM, which results in a MIMO channel.
In the field application of ADSL, the first generation of ADSL (or G.DMT) has only considered a downstream power cutback/politeness. It can be viewed as DSM Level 0, as it is merely to avoid signal saturation in the shortest loop lengths. The second generation of ADSL (ADSL2 and ADSL2plus) has DSM Level 1 consideration, as it provides both the upstream and downstream power cutback and can be decided by both central office and customer premises equipment (CPE) sides together. However, it only has one-way negotiation, meaning that if one side chooses larger power cut then it will be the final decision.
In the field application of VDSL2, it can be considered as DSM Level 2. The detailed power spectrum shapes of both upstream and downstream may be decided by CO side, with the negotiation of both CO and CPE. Later with the vectoring standard, VDSL2 also has DSM Level 3 implemented. A Vectoring Control Entity (VCE), residing at CO site, controls all connected CPEs with aligned symbol boundaries so that the desired signals and crosstalk are orthogonal, and crosstalk is cancelled by matrix operations. The overall of all users' rates are significantly improved by these DSM techniques. The high frequency bands suffering crosstalk among users are much improved by vectoring that cancels most of the crosstalk. This enables the overall average user data rates at least 95% of what they should get as if there exists no crosstalk interference. Comparably, without these techniques applied, the overall average user data rates may suffer 30˜50% degradations by mutual crosstalk among them.
Despite these prior art teachings, however, there exists an area that has not yet been considered. The DSM Level 2 considers the power back-off or PSD shaping strictly by the electrical length (the estimated loop distance between CO and CPE). For shorter loop distance, the PSD or power will tend to be lower, as it only needs less power to reach its service requirements; this power and power spectrum density (PSD) reduction also helps the overall system because its crosstalk into other users is also reduced. The final decision of power/PSD is at CO side, while CPE can only negotiate and suggest even lower power than CO side. If the line condition is not good, for example it has some static environment noise or radio frequency interference, the reduced power/PSD may leave it unable to reach its desired optimal rate. It may even be unable to reach its service rate.
An object of the present invention, therefore, is to provide an improved system and method for maintaining an optimal rate with balanced power and power spectrum density (PSD) reduction due to shorter distance and the consideration of its own noise floor.
Another object of the present invention is to provide an improved initialization exchange protocol, to facilitate the final decision of the power and PSD levels by considering both electrical length and the noise profile.
A related object of the present invention is to provide an improved system and method for balancing the far-end crosstalk (FEXT) of the overall system and each individual's noise characteristics.
A further object of the present invention is to apply the above methods in any high-speed DSL systems which may need the power reduction control for system FEXT performance, while also achieving individual DSL line's optimal performance.
A system of the present invention therefore eliminates the possible suboptimal rate degradation by the sole decision of power reductions by referencing the estimated electrical length between CO and CPE sides. In a preferred embodiment, the signals carry information in accordance with known protocols and standards. The system first includes a training protocol, which is used for the individual system to identify its characteristics, such as the loop distance, static environment noises, radio frequency interference, and so on. In the whole DSL system, the self-crosstalk is becoming more and more critical, and fortunately it can be greatly reduced by advanced DSM technologies. One important technique is to reduce the transmit powers or PSD levels for the lines with shorter distance between CO and CPE end terminals. Such technique is used for crosstalk mitigation for near-far problems; shorter distance means CPE is near CO and it creates stronger crosstalk to farther CPEs. These lines don't need the full powers or PSD levels to achieve their service rate, due to their much fewer signal attenuations. So-called power back-off (PBO) can be implemented by both ends' transmitters, with the knowledge of the estimated loop distance, or electrical length. Such PBO technique is effective to reduce the strong crosstalk (self-FEXT) from these shorter lines into other longer lines. Both CO and CPE end terminals measure their received signals, with the knowledge of the peer's transmitted PSD level, they further estimate the signal attenuations and the loop distance. Both CO and CPE end terminals also measure their noises, or the Signal-to-Noise-Ratio (SNR), to decide whether the PBO is able to prevent them to reach the target service rates.
Without the invention, the PBOs decided solely by the loop distance might cause the PSD to be too low, where the reduced signal levels compared to its noise levels are not sufficient to provide enough SNR to meet its service rates. This is not desired, as the quality of such short loops should be very sufficient to provide the needed service. Once they know the resulting SNR possibly fail to support their service rates, it requires a mechanism to adjust the PSD or PBO so that the newly transmitted signal has the desired level for the required SNR.
In this regard, one implementation adds an extra exchange phase, once the receivers gather the noise information and SNR, to adjust the peer's transmitter PSD if needed. In the VDSL standard, the protocol for the PSD/PBO decision is done in the Channel Discovery stage, which is the first stage of initialization. The signal measurement and noise measurement may be performed in this stage. But in the current protocol, the self-FEXT is present in this stage, which will mislead the noise measurement. The self-FEXT will be measured and cancelled in the second stage, Training & Analysis stage. Until then, the real noise measurement and resulting SNR are meaningful for final service. Therefore, the proposed implementation may include a retrain mechanism to restart a new initialization process so that the PSD/PBO decision may take the noise into consideration. This process may be optional if the measured noise will not affect its targeted service rates given the current PBO.
In the Channel Discovery stage, several messages exchange between CO and CPE ends. O-SIGNATURE is the first message in this stage. It conveys the CO's setting about PSD masks, UPBO (Upstream PBO) parameters, and many others. CPE may start measuring the signal in O-SIGNATURE; with the actual PSD information in this message that CO is sending, CPE may derive the channel attenuation and thus derive the loop distance/electrical length. The channel or loop attenuation is the signal gap between the transmitter's PSD and the receiver's PSD levels. The physical loop distance or electrical length is a single value kl0 representing the loop attenuation across the used bandwidths. There is a predefined rule to decide the PBO which involves UPBO parameters a and b, the electrical length kl0 and the subcarrier frequency. Next, CPE starts sending its first message, R-MSG1, with the actual PSD/UPBO it is applying. CO then in turn measure the signal, together with the PSD information in R-MSG1, it derives the channel attenuation and loop distance (or electrical length). CPE will also convey its estimated electrical length to CO side, and CO will make the final decision of electrical length in next message O-UPDATE. CO may assign a PSD ceiling to further limit the upstream PSD. The Upstream PBO (UPBO) is finalized by this final electrical length and shall be applied by CPE end in the beginning of Training stage. The Downstream PBO (DPBO) is finalized after receiving R-UPDATE, in which CPE may request a downstream PSD ceiling, and shall be applied by CO end in the beginning of Training stage as well. In the O-PRM, CO conveys the final decided PSD/DPBO to CPE end; likewise, in R-PRM CPE conveys the final decided PSD/UPBO to CO end.
As noted above, both sides shall measure the noises in additional to the signals. To better measure the real noises after self-FEXT is cancelled, this process may take place in the second stage of Training & Analysis. CO side will coordinate all the lines it connects to and try to cancel the FEXT to its best. Then both sides measure the real residual noises, and then decide whether the SNR is enough to support its target service rate. There can be two possibilities: (1) the SNR is enough, so the PBO/PSD levels are proper, and it proceeds to the final stage; or (2) the SNR is not enough, so the PBO/PSD levels need adjustment. In the latter case, a retrain may be needed to apply the adjustment since the PSD levels are finalized in the beginning of Training & Analysis stage. During the retrain, this adjustment may be implemented in the messages R-UPDATE and O-PRM. The R-UPDATE conveys the requested DPBO PSD upshift, in order to increase its received signal level and thus SNR. The O-PRM conveys the requested UPBO PSD upshift, to achieve the same purpose on the upstream direction. Both ends decide the final PBO PSD and apply them in the beginning of Training stage as mentioned.
Although the inventions are described below in a preferred embodiment involving a VDSL transceiver, it will be apparent to those skilled in the art the present invention would be beneficially used in many situations where it is necessary to reduce the risk of insufficient rates due to the power back-off.
An embodiment of a system 100 of the present invention is illustrated in
A preferred embodiment of a system 201 of the present invention is illustrated in
The present invention introduces a new examination on the SNR here, to ensure its target service rates will not be compromised by the PBO. Block 210 (i.e., retrain to set new PSD) is the additional stage if the PBO is not proper, it has to go back to Channel Discovery stage to negotiate the PBO again. A detailed negotiation will be explained in
Stage 205 is the third stage, Exchange. Both ends will finalize all remaining parameters and prepare for the entry of Showtime process. They will also exchange these parameters so that the peer may prepare its transmitter as well. If everything is fine, they enter stage 206, which is Showtime. At this point, the training and initialization are completed, and the data transfer and services may start. In the embodiment, any of the Channel Discovery stage, the Training & Analysis stage, and the Exchange stage can be an initialization stages of VDSL protocols performed by the system 100.
Ideally in Channel Discovery stage, all information including the signal, noise, attenuation, sender's PSD level etc can be gathered. However, the noise measured in this stage will not be final; the strong crosstalk in the DSL system will be handled in the next stage Training & Analysis. Once the pre-coder and post-coder are trained for the crosstalk, it is generally not desirable to adjust the PSD level again; in this regard, the standard does not allow any PSD change from this point. After the crosstalk is mostly cancelled, the real noise floor can then be used to estimate the final achievable rate. Once it decides whether the PSD level is too low or enough, it may decide the next step. If the PSD level is too low, it may trigger a retrain back to Channel Discovery stage. In there, they may propose the PSD upshifts in order to achieve higher rates. On the other hand, if it decides the PSD level is enough to support the target service rate, it may continue into the final stage and transition into Showtime.
Although the present invention has been described in terms of a preferred embodiment, it will be apparent to those skilled in the art that many alterations and modifications may be made to such embodiments without departing from the teachings of the present invention. For example, while the above description uses VDSL as an example, the teachings of this disclosure can also be applied to other DSL technologies, such as G.fast and other members of the family of technologies generally known as xDSL. It is noted that typically, the retrain portion of Block 210 may happen in later stages instead of the proposed Stage 304. For the Power/PSD level decision flow, the additions of message fields exchanged are possible to be appended into other messages, instead of the Message 404 and 405. The scope of the invention should, therefore, be determined not with reference to the above description, but instead should be determined with reference to the appended claims along with their full scope of equivalents.
