(CED Via Thomson Dialog NewsEdge)

Most industry veterans acknowledge that bandwidth has long been the advantage that cable has held over competitive access technologies. Hybrid fiber/coax (HFC) networks have been modified in myriad, creative ways to expand network capacity. The continuing evolution of HFC distribution systems has allowed the cable industry to offer incremental services horizontally (more video channels and types) as well as vertically (data and voice), while maintaining some amount of overhead for future services and applications.

Cable's unique ability to maintain the equilibrium between network capacity and bandwidth utilization is what has made this expansion not only possible, but fundable as well. HFC networks have proven to be exceedingly adaptive, but capacity always comes at a price, especially when bandwidth expansion efforts ripple across thousands of miles of physical plant, touching thousands of homes along the way.

Increasingly, the relationship between revenue, investment, and bandwidth utilization is eroding. In recent times, peer-to-peer (P2P) services (primarily audio file sharing services) have absorbed a disproportionate amount of network capacity on cable data networks with no corresponding revenue contribution. Dealing with this application put many operators in the awkward position of containing the "few" to better serve the "many," bolstered only by the fact that much of the P2P traffic is being used to support the illegal duplication of copyrighted material.

Unlike these first-generation P2P applications, a new breed of "over-the-top" (OTT) services will put cable operators in a more contentious position. Some distinguishing characteristics of these new OTT services are as follows:

Video as the predominant form of distributed content

Video content requires five to 50 times more bandwidth than audio files, affecting both the bandwidth and session length of each user sessionboth of which are critical factors in network traffic models and system design.

Lawful content distribution

The concept of lawful content distribution is becoming increasingly complex, with subscriber-activity largely based on the adage, "Possession is nine-tenths of the law."

Peer-to-peer seeding schemes for video content

P2P (used by some, but not all, second-generation OTT services) stands to dramatically impact traffic modeling metrics, all of which are designed to manage "peak" utilization periods.

In theory, cable operators have the tools at their disposal in the form of node splits, advanced modulation techniques (DOCSIS 2.0), and even channel bonding (DOCSIS 3.0) to address their capacity issues for years to come. The gap between theory and practice, however, is proving to be substantial, and the implementation of these advanced capabilities may be outpaced by evolving usage patterns and video-rich applications.

Conventional bandwidth expansion challengesUp to now, network capacity has effectively been able to outpace utilization, with bandwidth-per-user increasing at a rate of 10 percent to 30 percent per year. But what happens as network traffic becomes increasingly video-based, resulting in user sessions that consume exponentially more bandwidth for longer durations?

Node splitsthe first line of defense for expanding network capacitycan potentially cut serving area size in half, but this approach has complex network operational implications, not to mention significant construction costs and timing constraints. And evidence suggests that employing DOCSIS 2.0/advanced PHY techniques as a bandwidth expansion solution is equally unattractive. To date, no cable operator in North America has deployed DOCSIS 2.0 in a meaningful way, stemming from the complexities of intermingling different modulation schemes. Some suggest that DOCSIS 3.0 may represent a more viable solution than DOCSIS 2.0, but any next-generation solution is only effective to the extent that compatible consumer premises equipment (CPE) is deployed in the field, and no DOCSIS 3.0 CPE devices (based on a finalized CableLabs specification) are readily available today. As a practical matter, node splits and advanced modulation techniques have only been employed on a selective basis to solve "hot spot" problems. Network-wide upgrades that exceed the basic capabilities of DOCSIS 1.1 have not yet been justified on a risk/return basis.

To be fair, most broadband network traffic models were conceived based upon "pre-broadband" (dial-up) metrics, with a heavy dose of over-provisioning added to account for the unknown. As previously indicated, P2P file sharing applications have absorbed much of the initial over-provisioned capacity that had served as the levee for network capacity. Unfortunately, over-the-top video
content could represent, in aggregate, a network burden that is substantially greater than any other broadband application. As such, this new generation of OTT services can engender either unprecedented network contention or massive infrastructure costs with no offsetting increase in revenue
.

Regardless of cost, broadband service providers will likely be in the position of providing the "best for the most." One vehicle for providing such service lies within PacketCable Multimedia (PCMM) platforms, which allocate "Quality of Service" parameters such as bandwidth, latency, and jitter (video stability) for any selected application on a DOCSIS-based high-speed data network. Used properly, PCMM can create a kind of flood control system that manages fair use in times of congestion, but is otherwise inactive and unobtrusive.

Implementation guidelinesWhile proposed legislation and service provider sentiment certainly favor content prioritization, the actual implementation of a reasonable stratification system must be fair and rational. Our "inclusion-based" philosophy is derived from the following market generally-accepted usage data and factors:

~50% of peak broadband network traffic is simply Web surfing, but content is increasingly becoming video-oriented and bandwidth intensive.

~25% of peak broadband network traffic is used for instant messaging, although the user experience is largely "elastic"the user experience is largely unaffected by bandwidth, delay, and latency.

P2P traffic represents ~10%15% of peak network traffic, although the vast majority of these sessions are used for unlawful duplication of copyrighted material.

E-mail represents 10%15% of peak network traffic, but has varying levels of elasticity depending on whether it is associated with commercial or residential users.

Digital voice traffic, which represents a small-but-growing percentage of broadband network traffic, must be given priority to achieve appropriate levels of service.

Service implementationGiven that network contention will likely be unavoidable in the near future, we believe that the appropriate approach for service providers is to offer the best service for most subscribers during peak times, with the flexibility to re-allocate bandwidth during less contentious intervals. In essence, certain priority traffic must be statically provisioned for QoS (such as PacketCable 1.x voice), while other priority applications (e-mail) can be easily tagged for QoS by DPI platforms, or via other application-marking technologies. A more flexible, dynamic solution is required to address the increasing amount of streaming video content that is embedded in, or linked to, most Web portals, but the underlying structure of our inclusion-based bandwidth management system is depicted on the previous page.

Use casesClearly, every major Web portal is adding enhanced video distribution and search capabilities to its service in order to keep ahead of the burgeoning demand for media-rich content. While Web surfing is considered a utility function by most consumers, the fact remains that the load is placed on the broadband network. Regardless of whether the content is elastic or inelastic in nature, the broadband network is rapidly transitioning to a world of more concurrent users and longer online sessions. The challenge for cable operators going forward will be to maintain subscribers' perceived right to certain utility services while enabling differentiated capabilities that meet the demands of applications and users.

For the general surfer, adding QoS to the media-rich portal content makes optimal use of network capacity while preserving the crisp responsiveness that broadband users have come to expect. The diagram to the left depicts how such a capability can be readily enabled using a "Web Server Smart Agent," a modular extension to the PCMM platform that marks specific content on a portal or Web site.

For premium users, broadband network operators need to provide both content providers and consumers with the opportunity to accelerate the performance of their chosen application. Some content providers may choose the "pay for performance" construct promoted by several telecom executives. As a practical matter, the doctrine of fairness will likely require operators to offer QoS-enhanced capabilities beyond their own content portfolio. To complete the scenario, a manual turbo button-style burst of bandwidth can be used for any-and-all network traffic not already integrated into the PCMM back office infrastructure, giving consumers the ultimate control over the rate at which content is delivered to them.

SummaryWithout question, cable operators are now facing the challenge of managing over-the-top content and, in particular, the proliferation of video-over-IP. Finding the equilibrium point between customer satisfaction, network capacity and capital investment will be a delicate balance. Fortunately, PacketCable Multimedia systems, accompanied by complementary technologies like packet inspection platforms and smart agent subsystems, will give service providers the ability to manage their strained capacity in a manner that best serves their customers.

Email:
[email protected]