SAN JOSE, Calif. — Hockey scouts scour the world for talent for their teams to develop. Cisco System Inc.’s scouts keep an eye out for technologies that offer a business opportunity for the networking giant.
Stephan Monterde, Cisco’s senior manager of corporate development runs the company’s Technology Radar program. A worldwide network of scouts keep an ear to the ground and submits short summaries of technologies and trends they think will have an impact on Cisco’s business, he told the annual Cisco Editors Conference here. After a quarterly selection meeting and further research, an assessment panel maps the most significant trends by their relevance to the company.
The purpose is threefold, according to Monterde: to help commercialize academic technology and keep a pipeline to upcoming talent, to assess gaps in business strategy and to develop best practices to be shared with partners.
High Efficiency Video Codec (H265). Most of the video consumed over cable or satellite services uses the MPEG 2 codec, said Kip Compton, CTO of video collaboration with Cisco. The H265 codec can double the quality or the amount of video over the same bandwidth. “We think this will have a profound impact on mobile video,” said Compton, and he expects 70 per cent of all mobile traffic to be video in the next few years.
Ultra HD (4K video). With about 16 times the pixel information of standard HD, the video performance is “quite remarkable,” Compton said. Dovetailing nicely with developments in the H265 codec, Ultra HD will redefine what an immersive experience is, he said.
Web Real-Time Communication (RTC). Coming from the domain of the World Wide Web Consortium, Web RTC aims at turning every Web browser into a videconferencing endpoint. It’s predicated on the adoption of HTML 5, said Compton: “It’s embedded in the HTML 5 app.” This would make it easy, for example, for customers to interact with a company expert or representative within a Web page.
Real-Time Conversational Speech Recognition. Interactive voice recognition systems and transcription software have been around for years. “But even if they’re large vocabulary applications, they’re very constrained,” said Ananth Sankar, distinguished engineer with Cisco. These systems, including Apple Inc.’s Siri, are all applications involving a human speaking to a machine and trying to accomplish something. Natural conversation is a different animal. If a user doesn’t have to speak in stilted terms for the computer, conversations with friends and colleagues could be parsed, summarized and used to populate to-do lists, for example.
Real-Time Conversational Speech Translation. A step further down the road, real-time conversational speech translation is a three-step process: speech-to-text, text-to-text translation, and translated text-to-speech. The second stage is the bump in the road because of language barriers; for example, in the question “When does the product ship?” the word “ship” could mean “be sent” or “boat.”
In-Camera Analytics. In a retail environment, eight per cent of the people who enter a store buy something. The other 92 per cent? The store has no data on why they didn’t buy anything, said Guido Jouret, vice-president and general manager of video collaboration at Cisco. But improvements in surveillance cameras provide the ability to add software to do some analytic work on the captured video. In the retail example, analytics could map “hot spots” for traffic; if there’s a lot of traffic but not corresponding sales at a display, the price may be turning shoppers off. The technology naturally extends to a wide range of domains, from monitoring mentally ill in-patients to adjusting traffic light cycles according to the traffic observed.
Software-Defined Networking. Network devices that can be reconfigured on the fly by software are available, but there’s no ecosystem of applications that take advantage of the programmable interface, according to Dave Ward, vice-president of engineering, CTO and chief architect with Cisco. “That’s what the programmable interface and orchestration are going to allow us to build,” he said. Right now, information held within the network isn’t being extracted from the hardware, to allow, for example, tuning of bandwidth to control quality of service to thousands of mobile video clients at a time. Bi-directional extraction of data “is really what software-defined networking can be about,” Ward said.
Silicon Photonics. The merging of laser and photonic technologies is “absolutely, positively the most interesting thing going on” in the world of application-specific integrated circuits (ASIC), Ward said. Photonics allow chips to be denser, cheaper and lower-powered, and manufacturers can use different materials in their boards, he said.