Video on Demand: If You Build It, Will They Come?

Video on demand (VoD) is a commonly misunderstood term today. With real VoD, you can watch anything you want, anytime you choose, at an acceptable quality of service, even if that means Casablanca at 2:00am on a Tuesday morning. No firms currently deliver this capability in a consistent, high-quality manner that consumers deem acceptable. Research by Joakim Kalvenes and Neil Keon, professors at SMU’s Cox School of Business, shows how providers can use their existing infrastructure and resources to guarantee quality of service for VoD.

Quality of Service Remains the Big Hurdle

Keon notes that the only way for firms to guarantee quality of service between a back-end server and an Internet service provider (ISP) is to provision bandwidth themselves—whether they’re delivering video, video conferencing, voice, or music. He points at Vonage, the Internet phone company, as a good example of the problems service providers can experience. They occasionally have service interruptions because the backbone between two users’ ISPs is the Internet. When the traffic load on the Internet is light, the Vonage service works. But when there is a heavy load, it stops working. This same problem currently exists for companies striving to provide high-quality VoD. Until firms can publish a credible quality-of-service guarantee for VoD, consumers will remain skeptical. By studying and measuring video traffic, Kalvenes and Keon make a significant contribution toward creating a viable and scalable VoD service, with guaranteed quality of service.

Public vs. Private Networks

The Internet is considered a public network, while private networks are owned typically by phone and cable companies, such as AT&T and Comcast. Certain types of digital media are more suited to private networks, others to public networks. For example, for three years Victoria’s Secret attempted to air a program promoting its products over the Internet. The web proved not to be the ideal network for delivering this kind of content, as the webcast could not be adequately viewed by significant portions of the target audience. So Victoria’s Secret finally aired the program on CBS, the broadcast medium better suited this type of programming.

VoD works particularly well in niches such as high-end gated residential communities with more static networks. However, the authors have developed an analytical method that would enable providers to allocate their infrastructure resources according to prescribed calculations and produce quality assurance for VoD while complementing other digital offerings.

Today, phone and cable companies are best suited to deliver VoD because they own the fiber that the Internet runs on. By deploying their existing resources and properly managing their bandwidth, such companies could deliver reliable VoD, whether offering it as a standalone service or bundled with other offerings. The method proposed by Kalvenes and Keon could also be applied to video conferencing, gaming, and other digital activities.

Efficiencies Stakes

The key for companies to deliver VoD successfully lies in their ability to use the bandwidth of their existing infrastructures as efficiently as possible. Keon explains, “You can only cut out waste if you know precisely how much to cut. The people who provision the resources (bandwidth) at firms today have only a ‘ballpark’ idea about how much capacity can be squeezed through the network given certain traffic loads.” This research methodology ensures that by using the prescribed methods, quality of service can be guaranteed 99 percent of the time (an industry standard), or at whatever other level firms would tolerate in their business models. 

When phone companies design their telecommunications systems, they establish a performance standard to make sure their infrastructure can handle the worst traffic day of the year—Mother’s Day. In a similar manner, VoD needs to be built to deliver acceptable quality of service on Christmas day, for example. Kalvenes explains, “These systems must be built to withstand the worst-case scenarios, which we sampled for in a random selection of titles and start times for a vast array of system capacities. Compressed videos have variations that can only be described randomly. Our research uses statistical methods that consider all probabilities.” For example, The Godfather didn’t work very well due to its encoding. Toy Story ran beautifully, most likely because it was produced digitally with compression in mind. A system would need to account for numbers of viewers simultaneously watching The Godfather, Toy Story, and other titles, within the bounds of probability.

The gains from efficiency (e.g., not allocating unnecessary bandwidth) can be consequential. Additionally, the authors’ methodology can adapt to unanticipated increases in demand in real time as well as instances when cable or phone companies change the topology of their distribution networks. The formulas would still derive the proper allocation of resources.

Conclusion

VoD is fundamentally different than broadcast or other video presentation methods, which can be anticipated. True VoD is demand that cannot be anticipated, and as such it has proven difficult to deliver in a reliable fashion. SMU Cox Professors Joakim Kalvenes and Neil Keon have developed a methodology that can enable providers to use their existing infrastructure and resources to deliver VoD with guaranteed quality of service.

The research paper “Traffic Estimation and Capacity Assignment in Multimedia Distribution Networks with Guaranteed Quality” by Joakim Kalvenes and Neil Keon is currently under review.

Research Summary by Jennifer Warren