The Traffic Flow Managing

Highlights :

You are here: Home

Engineers

Electronic Engineers

The Traffic Flow Managing

Written by Eng.Admin

Friday, 07 November 2008 11:55

SCTE Cable-Tec Expo 2008 provided valuable nuggets of information regarding traffic management, the transition of the cable network to IP, deflecting competitive heat, and the digital transition.

While not as ubiquitous as edge QAMs were at last year’s SCTE Cable-Tec Expo, Internet traffic management emerged as a prominent theme in booths and in sessions at the show in Philadelphia.

During a morning breakfast session, moderator Ken Wright, Arris’ CTO, asked a panel that consisted of John Schanz, Comcast EVP, national engineering and technical operations; Liberty Global CTO Balan Nair; Time Warner Cable EVP, advanced engineering, Mike Hayashi; and Rogers SVP, network engineering and operations, Dermot O’Carroll about traffic management.

While network management has garnered a lot of attention this year, specifically with Comcast, O’Carroll said it has always been a service provider’s duty to make sure all of its customers get the best Internet experience possible. In general, the panel said cable needs to get away from managing customer usage through protocols that target specific types of users, such as peer-to-peer.

Four months ago, Rogers started a program that places caps on the amount of Internet usage by subscribers, and charges extra for those who go over their limit. Rogers set up a page on its Web portal that allows subscribers to check where their usage stands. Customers who are over the limit are shown what their extra-costs would be for going over the cap, but Rogers “zeroed out” the additional charges. That changed on the next billing cycle after Expo, when customers were charged for exceeding their bandwidth limits.

While Rogers benefits from Bell Canada using a similar strategy, Arris’ Wright said consumption-based billing south of the Canadian border could be risky for cable operators in the United States because it could drive affected subscribers into the arms of telco competitors.

Figure 1: The move to IP is encouraging the consolidation of headends. With IP transport, it’s easy for headends to back each other up. Furthermore, bandwidth in the regional network is relatively cheap versus bandwidth on trunk lines. Source: Cisco Systems, “IP for Video Heads” by Dave Brown.

Schanz said Comcast’s publicized run-ins with peer-to-peer companies, some of whom Comcast is now working with to find better solutions, have raised the public’s awareness about network traffic management. Time Warner Cable’s Hayashi said the cable industry needs to do more to educate consumers and bloggers about traffic management because “awareness is critical.”

While not naming names, Comcast’s Schanz said during the panel that his company is testing several vendor platforms to find better approaches for congestion management. Schanz said Comcast hopes to be finished with its trials by the end of this year.

Certainly Arris and Camiant are two companies that come to mind when traffic management is mentioned, and both had demonstrations at Expo.

Camiant has applied its policy control and application assurance technology to its Fair Use Management (FUM) application, which made its debut at The Cable Show earlier this year.

Randy Fuller, Camiant’s vice president, business development, said the application, which works with Camiant’s policy server, applies sophisticated policy management techniques to establish and enforce bandwidth quotas.

The policy server can be used to provide notifications to either cable operators or subscribers on bandwidth usage. By gathering information from cable modem termination systems (CMTSs), the policy server and Fair Use Management application are able to set policies to dynamically maintain balanced broadband performance no matter what the sources of high consumption might be at any given time within a service area.

While being application agnostic, Fuller said the system is able to “tamp down” a heavy user during times of heavy congestion on a network.

Camiant can also help on the consumption-based billing side by letting subscribers know when they’re getting close to their bandwidth quotas, and also aid them in upgrading to higher tiers.

Arris was demonstrating its Fair Bandwidth Manager (FBM) at Expo. FBM combines the edge monitoring of Arris’ ServAssure Advanced platform, which came from the C-Cor purchase last year, with the on-demand capabilities of the company’s Policy Service Manager when Internet traffic needs to be managed.

CTO priorities: The technology chiefs of the major MSOs have a lot on their plates: analog reclamation, HD, network convergence, tru2way, DOCSIS 3.0, and more, but there’s no industry-wide consensus on how to prioritize the deployment of those technologies. The panel included, from left to right, SCTE’s Dr. Roger Blakeway, Cox Communications’ Chris Bowick, PBS’ John McCoskey, Comcast’s Tony Werner and HBO’s Bob Zitter. The panel was moderated by Leslie Ellis.

The Fair Bandwidth Management platform helps operators identify specific customer data usage levels that are causing the congestion and, during peak usage times, adjust those specific users’ speeds without disrupting their quality of experience, while alleviating the congestion affecting all other customers online. Arris said it can improve customer satisfaction for more than 90 percent of the subscribers on a congested interface.

Wright said the FBM looks at each service group or node to find potential problems on the devices and then compares the usage with the service level agreements of the various subscribers. FBM is a service protocol agnostic tool that applies “throttling gates” in a CMTS once the problem devices have been identified.

When congestion starts on an interface, ServAssure sends an SNMP “high utilization” trap to FBM. FBM then sends a query to the ServAssure Advanced data warehouse to find out the utilization and service class for each device on that interface.

Using the service class information to distinguish premium or platinum level users, the FBM then applies a multiple optimization program (MOP) algorithm to determine which devices need to be managed and the minimum amount of intervention needed to relieve the congestion. Once the devices are identified, the ServAssure Advanced resolves the IP addresses of the devices to manage, and then the throttling gates are applied to the appropriate CMTS to temporarily lower the bandwidth to the problem IP addresses that were identified by the MOP.

As soon as a “congestion clear” trap is received, the throttling gates are removed and full bandwidth is restored to the heavy users.

RFMD
RF Micro Devices (RFMD) was a new, old face at this year’s Expo. While RFMD was making its Expo debut in Philadelphia, in 2007, it purchased Sirenza Microdevices, which was at last year’s Expo. RFMD was able to combine the hybrid amplifiers from Sirenza with other products in its portfolio, with the end result being the introduction of what it says is the industry’s first gallium nitrate (GaN) based cable TV amplifier modules.

The modules, which are called the D10040200PH1 and D10040230PH1, were designed for use as power doubler amplifiers in current and next-generation cable TV infrastructure applications, including line amplification and hybrid fiber/coaxial (HFC) optical nodes.

RFMD said the D10040200PH1 and D10040230PH1 feature 20 dB and 23 dB of gain, respectively, and are the first power doubler amplifier modules.

As an example, cable TV infrastructure designers can use either the D10040200PH1 or D10040230PH1 in HFC optical nodes to increase final power output by 2 to 3 dBmV over competing devices, while maintaining equivalent multi-carrier distortion performance, in particular Carrier to Intermodulation Noise (CIN) levels.

Conrad Young, RFMD’s director of business development, Broadband and Consumer Business Unit, said the company got an early jump on GaN in 2001 when it started making GaN-related devices for the aerospace and defense industries. Aside of the GaN element, Young said the key to the hybrid amplifiers was their ability to integrate with other vendors’ products.

The hybrid amps are currently in production, with orders shipping overseas and qualification underway with a large vendor here in the United States.

Workshops & sessions
Tech gurus presented papers on a multitude of topics at Expo. In one example, there’s disagreement among MSOs over whether they’ll eventually migrate fully to IPTV – which is to say, whether or not to migrate to a fully converged network, according to Cisco Cable Solutions Manager Dave Brown, in a paper that chronicles the ongoing transition of the cable network to IP.

Brown notes that from a network perspective, transmitting IP video is relatively simple. The main impediment to an all-IP network is the large installed base of set-top boxes that can handle MPEG video but are incapable of processing IP video. Populating the field with hybrid boxes is one way to ease a transition, Brown suggests.

The industry is currently split. On the one hand, he said, are those who believe there’s nothing to be gained by encapsulating video in IP and transporting all video on the DOCSIS channel – so why do it?

On the other are those who believe it’s inevitable that all services will migrate to IP, because of the operational simplicity of a converged network and the bandwidth-saving benefits of easy statistical multiplexing and the use of variable bit rate (VBR) transmission.

Which view wins? “Time will tell,” Brown says. While time is telling, Verizon is deploying fiber-to-the-home. Consultant Victor R. Blake contributed a paper that notes that Verizon could honestly offer a 236 Mbps data service using the same oversubscription ratio cable is currently using. Even with DOCSIS 3.0, cable would have to split nodes to 10 customers a node to match that.

Figure 2: Three CMTS architectures: In addition to the i-CMTS, which decouples downstream channels and upstream channels, and the m-CMTS, which separates the downstream PHY functions of the CMTS and moves them into edge QAMs, a third potential architecture has been proposed. The partitioned CMTS, or p-CMTS, would also put downstream PHY in edge QAMs, but it would also separate out upstream PHY and place it… someplace to be determined. Source: Arris, “On the Evolution of the HFC Network,” by Tom Cloonan.

Bit of a scare tactic, that. On the other hand, Blake notes, Verizon isn’t offering 236 Mbps, and DOCSIS 3.0 will be more than adequate for now. And if Verizon does start dramatically upping the transmission rates it offers, then cable can respond with DOCSIS over Ethernet PON (EPON), which Blake evaluated as requiring both lower capex to deploy and lower opex to operate. It can be rolled out gradually, as needed.

On the other hand, given that DOCSIS technology generations tend to last for about four years, after DOCSIS 3.0, the cable industry is likely to have ready some new version of DOCSIS that will probably be more than DOCSIS 3.0 but less than DOCSIS 4.0. Whatever its number, it will probably cover the years 2012 to 2016, according to a paper delivered by Tom Cloonan, chief strategy officer at Arris.

The issue for operators is making maximum use of the resource of bandwidth. Cloonan surveys the current options, and tallies 17 – 10 of those make more efficient use of existing bandwidth (e.g., expanding to 1 GHz, splitting nodes, using stat muxing and VBR), three that augment but don’t replace extant coax (e.g. spectral overlay), and another four that replace coax with other solutions (e.g., RF over glass, PON).

How much bandwidth will each subscriber need? Cloonan considers several scenarios, and calculates that operators should be able to support nodes of about 250 homes each, with about 5 Gbps per node, at least through 2016. Upstream traffic might grow to be about 88 Mbps per node of upstream traffic.

In other words, there’s no critical impetus to migrate away from the familiar HFC network any time soon, although fiber-deep techniques and ongoing node splitting are assumed.

Cloonan said that CMTS companies are now doing preliminary work on designing CMTSs that will be deployed in 2016. Vendors might provide anywhere from 10 Gbps to 50 Gbps per CMTS linecard (and possibly up to 100 Gbps). Even a moderately-sized headend (serving about 40,000 homes) could be served by a single CMTS chassis – probably one that isn’t even filled.

There are three possible forms for the CMTS of the future: the i-CMTS, the m-CMTS, and the p-CMTS.

The i-CMTS is closest to the typical CMTS, but it decouples downstream channels and upstream channels, so that operators are not forced into a fixed ratio of downstream to upstream channels.

The m-CMTS separates the downstream PHY functions of the CMTS and moves them into edge QAMs.

The partitioned CMTS, or p-CMTS, is a fairly new proposal still being discussed. It suggests partitioning the MAC and PHY functions into separate chassis. The p-CMTS would also put downstream PHY in edge QAMs, but it would also separate out upstream PHY and place it… someplace to be determined.

Costs could be reduced significantly. PHY resources in the edge QAM could be shared between DOCSIS traffic and video. Cloonan evaluates the advantages and disadvantages of each in turn.

Some of you may have been asking yourselves: “What is TISPAN, and what does it have to do with PacketCable 2.0?”

Glad you asked. Camiant Marketing VP Richard Cardone explains TISPAN (an acronym tortured out of the phrase Telecoms & Internet converged Services and Protocols for Advanced Networks) is a set of IMS-based standards being developed for the wireline industry, as PacketCable was designed for the cable industry.

CedarPoint Senior Director Peter Quigley explains that both TISPAN and PacketCable 2.0 extend IMS (IP Multimedia System) standards. PacketCable 2.0 and the TISPAN both add standards to IMS that cover simulation/emulation of the public switched telephone network (PSTN).

PacketCable 2.0 is established, while TISPAN is still evolving. Though rooted in Europe, Quigley suggests TISPAN is something the U.S. cable industry should keep an eye on.

The digital transition may be something broadcasters have to deal with, but it will not be a simple issue for cable, which plans to keep providing analog signals for at least three years after the transition. Matthew Goldman, VP of technology at Ericsson’s Tandberg Television operation, warns that downconverting from HD into AD requires more than just a video format translation.

MSOs will also have to take into account audio dynamic range compression, loudness adjustment, the formatting of closed captioning and content advisories, and the extraction and transmission of Nielsen audience measurement and TV Guide data.

transfered

Comments

Add New