April 1, 2010

Reality Check: Upstream: What’s Next?

For years, cable operators have focused largely on the downstream, and for good reason.

The industry has traditionally obtained most of its revenue from video services. Video is highly asymmetrical, using prodigious amounts of spectrum for downstream services and very little for upstream.

Cable networks are highly asymmetric by design. In the U.S., the current upstream encompasses 5-42 MHz. However, ingress and various forms of noise make the lowest end of the spectrum difficult to use. The band useable for upstream DOCSIS carriers is generally 18-42 MHz, and in some cases 22-42 MHz.

Thus: Out of 37 MHz total available spectrum in the upstream, only about 24 MHz is usable, compared to the 496 to 958 MHz available in the downstream.

Two main approaches for enhancing cable plants have been converting analog channels to digital and extending the downstream frequency band to 1 GHz. Switched digital video is another.

"Continue segmenting, and you eventually have a passive coax network."

All of these technologies increase the capacity of the downstream, but do nothing to enhance the upstream. It is time to focus on this increasingly precious commodity.

Many in the industry have helped define applicable solutions. Synchronous code division multiple access (S-CDMA) in the low frequency end of the upstream allows use of previously unusable return frequencies. The Next Generation Network Architecture (NGNA) project four years ago included a proposal for extending the return band to 5-85 MHz.

All HFC equipment suppliers, including Motorola, Cisco, Arris, Aurora, and Harmonic, support segmentation. In addition, Motorola has developed a scheme to use frequency upconversion deep in the RF plant to segment the upstream. Javelin Innovations has proposed creating new downstream and upstream bands above the existing downstream.

Segmenting the upstream and using SCDMA upstream do increase the available spectrum. These techniques should be implemented as soon as required, which in many systems is now. However, SCDMA provides only a modest, if needed boost.

Other schemes require changing the frequency allocation. This is expensive, and in cases where the manufacturer of the deployed equipment no longer exists, it may be impossible to implement.

Segmentation clearly can enable expanded upstream capacity. What is interesting is that if the network is segmented to a degree that allows for a passive coaxial plant, changes to frequency allocation become not only feasible, but also very practical.

Here’s a key detail: In most systems, active devices alone use diplex filters to split the downstream and upstream. Such filters make changing frequency allocation at best expensive and at worst impractical because every active device must be changed simultaneously. (Note that some operators use reverse attenuators or equalizers in the drops or taps, which will complicate changing the frequency split.)

The only filters in a passive coax RF plant are in the optical node. Therefore, changing the downstream/upstream frequency allocation typically requires updating only a single device.

This puts the industry in an enviable position. Continue segmenting, and you eventually have a passive coax network. Operators then can consider other cost-effective technologies to increase capacity.

Indeed, a passive coax network can match the throughput of a fiber-to-the-home (FTTH) network at a fraction of the cost.

By making several new technologies feasible, passive coax architectures enable significant increase in the available bandwidth, including upstream. Given the sunk costs, it makes sense to avoid undoing previous upgrades. The best way to accomplish this is for the industry to design segmentation upgrades with the goal of creating a passive coax plant.

-Eric Schweitzer has held senior technology management positions on the MSO and vendor sides of the industry.