ET/NCTA Technical Snapshots
There are several ways to slice and dice a trade show. This postmortem covers about half of the technical papers presented during the first-ever, three-day "Cable Connection, Spring" event.
Another first: A single committee evaluated all papers submitted to the SCTE Conference on Emerging Technologies and the NCTA Cable Show. The group apportioned 17 as NCTA Technical Papers, which were presented over five sessions in two days, and 18 papers to the one-day ET track. This review focuses on the ET papers.
For the NCTA technical paper titles and authors, see the sidebar.
At this year’s ET conference, the emphasis was on "emerging." DOCSIS 3.0 is clearly still emerging, and advanced high-speed data technologies and apps were featured in many papers. In contrast to earlier years where "emerging" meant "new," the theme this year was on emerging as unfolding.
Such unfolding takes place within a dynamic equation. Demands on high-speed data networks, for instance, continue to increase because of a variety of factors, including increased use of streaming media.
Online viewing in November 2008 jumped 34 percent over the previous year, while in one year, the average bit rate of one service provider grew by six times, Carol Ansley of ARRIS wrote in one of the papers presented in the first ET panel. That session was devoted to capacity planning and management in the downstream.
Cable operators should understand network activity using tools such as IP data metrics and DPI and should try to open dialog with Internet video providers through neutral organizations so they are prepared for what might be coming, Ansley wrote.
On the enterprise side, business customers use one Ethernet connection for several critical applications. Hierarchical QoS (H-QoS) helps maintain bandwidth efficiency and QoS on a per-application basis, Scott Sumner of Accedian Networks, wrote in a paper devoted to cable business services. Dedicated silicon-based architectures will increase the benefits to H-QoS even further.
Within the app
The Internet Engineering Task Force is developing specifications to allow applications themselves to help manage bandwidth, Richard Woundy of Comcast wrote in "Application-Centric Bandwidth Management and the IETF."
With application-layer traffic optimization (ALTO), an ISP can help a client using a P2P application select peers geographically. Another spec, low extra delay background transport (LEDBAT), uses a congestion control algorithm to delay "insensitive" application traffic. However, neither will replace the need for operators to manage their networks, Woundy wrote.
Adding spectrum and increasing its efficiency are "major" investments that result in "large" returns, Robert Howald of Motorola wrote in "Fueling the Coaxial Last Mile."
Pushing fiber deeper leads to bandwidth segmentation gains, cleaner spectrum and new spectrum. Specifically, fewer amplifiers in cascade means less noise and distortion, and cleaner spectrum results in greater bandwidth efficiency, Howald wrote.
Picking up here (and more so in his NCTA paper) where he left off in his 2002 Cable-Tec Expo paper subtitled, "Modulation to the Power of Ten," Howald includes 1,024-QAM as one component to reaching certain service requirement goals.
Yet, managing bandwidth is still cheaper than adding bandwidth, Alon Bernstein of Cisco wrote in "Bandwidth Management: a Multi-Layered Approach to Efficient Bandwidth Allocation" that operators should evaluate QoS tools based on time scale, whether they limit or manage congestion, network location, and whether they control bottleneck links.
Heavy on the upstream
As is clear from the session on maximizing upstream capacity, DOCSIS 3.0 poses challenges. These range from scheduler challenges to IPTV to everything in between.
As Bernstein, in a second paper, points out in "Simpler, Faster, Cheaper Upstream Scheduler," the scheduler required for DOCSIS 3.0 is anything but fast, cheap or simple. As a result of the bonding techniques used, requirements for backward compatibility and various service overlays, DOCSIS 3.0 features the most complex scheduler of any access technology in use today.
While a challenge for vendors, it is not cost-free for operators, either. The CPU resources and complex software will result not only in higher capital cost, but also greater technology risk. Bernstein’s analysis shows that it isn’t the service capabilities that are driving the complexity; it’s the overlay of those service complexities onto independent upstream and downstream channel bonding that drives the complexity.
With DOCSIS 3.0 being rolled out today, it remains to be seen how that complexity will affect long-term cost reductions.
Also in this session, Ayham Al-Banna and Tom Cloonan of ARRIS presented their analysis of "Performance of Multi-Carrier Systems when Applied to HFC Networks." Among other things, they project the lifetime of wideband cable modems until 2016 with a peak data rate of 288 Mbps.
The objective of the authors is to propose new areas for future PHY development to increase signaling efficiency. The same two authors, along with ARRIS’ Jim Allen, offered another paper titled "DOCSIS 3.0 Upstream Channel Bonding: Performance Analysis in the Presence of HFC Noise."
The ARRIS teams’ calculations of upstream bandwidth as 30 Mbps per upstream channel assume 64-QAM operation for 6.4 MHz channels. But that’s a far cry from today’s deployments where most 3.2 MHz upstream channels still operate at 16-QAM (and a significant number still at QPSK). As they point out, closing that gap will involve significant work in tightening the plant.
Their ultimate conclusion: "Finding the optimal settings for all of these knobs (PHY settings) was a time consuming task that could only be done in an automated fashion." In short, the CMTS and modem systems will increasingly be relied upon to optimize the PHY rather than manually configuring options. Operators and systems engineers will certainly welcome this ongoing trend. The integrity of the coax plant and quality of components will remain a challenge that will limit the performance of the networks – an issue not addressed in this paper.
The bit-squeezing theme continued with Motorola’s John Ulm on "Upstream Bandwidth Session Leveraging S-CDMA for Cost Efficient Upstream Capacity." Again, the focus is on upstream bandwidth. The authors argue that the "user’s perception of the differences between service rates becomes less apparent as data rates increase."
While that’s true for services such as email and Web browsing, it’s hard to extend that logic to streaming video and file sharing. The peak-to-mean ratio shows a greater spread as the peaks increase and mean consumption rises more steadily. This statistical fact is the foundation of the shared wideband architecture of DOCSIS 3.0.
A paper by Comcast’s Dave Urban on upstream capacity shares the same theme. Considerations for real-world problems, such as supporting legacy 1.0 and 1.1 modems, make this paper strongly recommended for operators considering a variety of architectures and spectrum planning (including non-DOCSIS applications). Although his focus is not on automation, the wide array of configurations he considers does demonstrate that automated solutions, as suggested by the ARRIS team, could lighten the burden for operators.
Collectively, the authors of these papers suggest improvements for DOCSIS, many of which fall outside of 3.0’s scope. Yet, CableLabs indicates the spec is closed to new features in order to provide stability.
Cable operators need ways to expand their infrastructure rapidly and cost-effectively to meet consumer demand for new services and stay competitive. As indicated by papers in a third ET session on scaling network infrastructure, newer technologies and architectures could provide relief.
Modular-CMTS (M-CMTS) could be a way to lower DOCSIS-transport costs while increasing capacity, according to Pawel Sowinski of Cisco. However, current M-CMTS specs do not address redundancy.
In "Resilient M-CMTS Architectures," Sowinski said the M-CMTS downstream external Phy interface (DEPI) connection is the "weak point" in the M-CMTS architecture. He laid out a proposal for redundancy that includes independent and redundant L2TPv3 tunnels.
M-CMTS implementation could bump up demand for QAM channels, as could more digital programming and an increase in narrowcast services. This growth is stressing cable operators, according to Jorge Salinger of Comcast.
In his paper, Salinger proposed a Next Generation Access Architecture (NGAA) to "consolidate" packet processing from CMTS and edge QAM, simplify port sharing between DOCSIS and MPEG IS, introduce a packet processing shelf, and put upstream and downstream MAC and PHY in the narrowcast platform.
While Ethernet PON (EPON) could help with fiber-to-the-premises connectivity, replicating DOCSIS back-components is cost-prohibitive. In "DOCSIS Management for EPON Networks," James Chenv of Salira Systems explained how DOCSIS over EPON architectures include a middleware layer that makes DOCSIS management instructions readable by the fiber network.
As for the transition to FTTH, Oleh Sniezko of Aurora Networks said RF over glass (RFoG) can ease the process without replacing CPE. His paper, "RFoG: How to Make it Work and How to Expand it," said reverse data speed capacity in RFoG is equivalent to xPON, thus solving the problem of bottlenecks in the upstream.
The IP future
Subs increasingly are watching Internet video and content on mobile devices or PCs. Cable operators can use IP-delivered video to meet both demands.
Xiaomei Liu of Cisco noted in the fourth, IP-focused session at ET that while content may travel from regional networks to hubs while encapsulated in IP, a QAM device generally has been used for MPEG conversion before traversing the last mile. DOCSIS 3.0 and M-CMTS standards provide cable operators with the mechanisms for carrying IP to the subscriber’s home.
Specifically, DOCSIS 3.0 downstream channel bonding and multicast allow operators to scale the network cost-effectively and bandwidth efficiently, wrote Sangeeta Ramakrishnan of Cisco in a separate paper.
Moreover, Ramakrishnan showed that IPTV requires a "fraction" of the 100 downstream RF channels used by current video services. Spectrum can come from analog reclamation and from channels used today for VOD and SDV, he explained, in "Scaling the DOCSIS Network for IPTV."
IPTV also improves efficiency by 40-60 percent with VBR and offers switching benefits comparable to SDV, Liu wrote. He examined migration strategies including an IP overlay approach and a hybrid yield approach for CPE, in "Making IPTV Make Sense in a Cable Environment."
Weidong Mao of Comcast presented several high-level architectures and sub-system and data plane/control plane interfaces in "Key Architecture and Interface Options for IP Video over Cable," and provided criteria for considering each. It is important to analyze whether the interface or architecture is standards-based, offers low complexity, high bandwidth efficiency, scalability, and low costs, and shares infrastructure with traditional video, he said.
Christopher Albano of Comcast examined how IP encapsulation of QAM-delivered content can be used to enable delivery to devices not installed by the service provider. It can be delivered using MoCA and Wi-Fi, while a DLNA-managed home network and centralized point of entry into a subscriber’s home can help simplify what might otherwise be complex handoffs.
Jonathan Tombes is editor of Communications Technology. Monta Monaco Hernon is a contributor. Victor Blake is an independent consultant.
Sidebar: 2009 NCTA Technical Papers
Quality Matters: Assuring Performance
• Michael Adams, Tandberg Television, "Content Management Systems vs. Content Delivery Networks"
• Dave Higgins, et al., Comcast Media Center, "Developing a Grading System for Digital Video Quality"
• Asha Kalyur, Cisco, "Video Quality Assurance across IP Networks, "Challenges, Requirements and Winning Strategies"
• S.V. Vasudevan, et al., Cisco, "The Headend Revisited: A Multi-Service Video Data Center for the Modern MSO"
Highly Individual: Digital Content Protection and Delivery
• Kevin Bauer, et al., University of Colorado, "The Challenges of Stopping Illegal Peer-to-Peer File Sharing"
• John Chapman, et al. Cisco, "TelePresence over DOCSIS"
• Andrew Poole, et al., ARRIS, "My (TV) Space – Using Subscriber Management Systems to Refine and Redefine Video-based Social Networking"
• Niels Thorwirth, Verimatrix, "A New Concept for Robust Video Marking"
Sending the Right Message: Targeted Advertising
• John Callahan, et al., Active Video Networks, "Cloud Computing for the TV: Expanding Choice, Control and Content for a New Generation of Television Viewer"
• Dan Holden, Comcast Media Center, "Increasing MSO Advertising Revenues through Management of Ad Skipping"
• Arthur Orduna, Canoe Ventures, "Innovation in Multi-Platform, Multi-Operator Advertising"
• Mitch Weinraub, et al., Comcast Media Center, "Connecting Cable’s Network and Applications with its Industry Partners while Maintaining our Individuality"
Scaling for Tomorrow: New Approaches to Video Encoding
• Ron Gutman, et al., Imagine Communications, "Separation of Video Processing and Multiplexing: The Key to Network Scaling of HD-VOD and other Bandwidth-Intensive Personalized TV Services"
• Robert Howald, et al., Motorola, "New Megabits, Same Megahertz: Plant Evolution Dividends"
• Dirk Jaeger, et al., Technical University Braunschweig, "DVB-C2 – The Second Generation Transmission Technology for Broadband Cable"
• Robert Thompson, et al., Motorola, "Optimizing Upstream Throughput Using Equalization Coefficient Analysis"
The Invisible Network: Wireless Broadband
• Timothy Burke, Liberty Global, "Wireless and Home Networking: A Foundation for Service Provider Applications"
• Tom duBreuil, Motorola, "An Overview of TV White Space for the Cable Industry"
• David Park, BelAir Networks, "Getting to 4G via Wi-Fi in the Real World"