The ABCs of Connectivity: Always Best Connected

1.9.03 -- The proliferation of cell phone service providers, with their dizzying array of caveats, costs, and coverage, and the looming convergence - as "standard" - between telephony (voice) and the Internet (mostly data) has caused the consumer no end of heartburn.

Do I have to carry multiple devices depending on the scope of my needs? Do I have to subscribe to multiple plans - not to mention remembering the varying particulars of each - to support them all?

Well, yes - for now.

Wouldn't it be nice to have one device - better yet supported by one seamless infrastructure - that could do it all, everywhere, at the fastest speed possible, for a reasonable initial investment and monthly cost, that didn't require a rocket science degree to learn how to use, and that didn't become obsolete in less than a year?

Rajesh Mishra and
(From left to right) Rajesh Mishra and
Per Johansson

Enter a research group at UCSD affiliated with Calit² doing its part to address the "seamless infrastructure" part of this problem. Their project is called "Always Best Connected."

While it requires a lot of technical terms and acronyms to explain, the concept is simple: Seamless interoperability among various network infrastructures to maintain constant - and the most powerful - connection to the Internet for mobile users using both licensed (e.g., cellular networks) and unlicensed (e.g., IEEE 802.11b) spectrum. This work is making it possible for user devices to seamlessly roam among varying network access technologies, including IEEE 802.11b, CDMA2000, GPRS, and Ethernet, with controlled adaptation of applications' performance to the network capabilities at hand.

802.11b is a wireless networking protocol that not only is spreading across both the UCSD and UCI campuses but is beginning to be popularized at retail outlets such as Starbuck's and, just a month or so ago, even a high-tech food court in the Mira Mesa area of San Diego. CDMA2000, associated with QUALCOMM and as the competitor to GSM, comprehends a family of standards that specifies a spread spectrum radio interface that uses Code Division Multiple Access to address the requirements for 3G wireless communication systems. GPRS, for General Packet Radio Service, is the data extension of the GSM voice service popular in Europe. And Ethernet is a local-area network technology ranging in speeds from 10 Mbps to 1 Gbps.

It's obvious that interoperability would make users' lives better. But what are the questions that motivate researchers?

"Our goals," says Per Johansson, senior researcher at Ericsson, one of the industrial partners in the project and a founding partner of Calit², "are to support user mobility and session continuity so as to maintain the application's state. This project is really all about what the applications need, and they of course vary a lot from one application to the next."

Video
Video of Larry Smarr playing at left. On right is user interface to the ABC system.

One of the challenges in transitioning between these access technologies is the dramatic difference in their capacities from an application's perspective. For example, in moving from a WLAN (IEEE 802.11b) to cellular access (such as CDMA20001xRTT or GPRS), the capacity goes down, causing the application to have to adapt - and quickly - so as not to adversely affect the user. This becomes a particularly serious problem if the application is multimedia-based, with a heavy reliance on voice and/or video.

"We're developing a software, rather than a hardware, solution," says Rajesh Mishra, research engineer of Ericsson. "Our company is interested in this project because it joins the services from WLAN, CDMA2000, GPRS, and potentially the Universal Mobile Telecommunications System, also known as WCDMA. Our goal is to find a way to integrate them so that users may choose any combination of the technologies according to their individual preferences."

Continues Mishra, "We're trying to enable an environment not only to prove the concept of interoperability but to provide a testbed for researchers - in both Electrical and Computer Engineering as well as applications areas - to conduct their own experiments."

"We need more operator involvement, not just to gain access to their commercial services," Johansson says with a smile, "but to get them involved in the research. That way we could analyze latency, network performance, and the effect of varying user loads."

ABC Operators, of course, make money by charging for time over the air. That's why they spend so much on licensing spectrum. Unlicensed spectrum, though, cannot generate revenue, so operators have been inclined to see hybrid infrastructures, such as ABC, as threatening to their business plan and potential sinkholes for technical support.

"Part of the problem," says Johansson, "is that operators need to figure out how to use and support hybrid schemes without becoming embroiled in a constant game of finger pointing when things don't work. This will probably cause them to change their business model to focus less on 'bits' and more on providing service. They need to realize that by giving [up] something, they'll get even more in the end."

Johansson points to voice mail on cell phones as a good example of companies giving something up to achieve a larger goal: "Voice mail is normally a free service. Why is it free? Because its availability encourages people to call more."

Besides operators, ABC is also interested in engaging people that want to develop applications. "Sixth College students with their PDAs are an ideal group for this project," says Johansson. "While these students are not all studying engineering, they are all being encouraged to develop applications based on the latest technologies. What we're curious about is what kind of applications they will develop when they're no longer constrained by the current WLAN coverage but, instead, are always best connected anywhere nationwide."

Q&A
Ramesh Rao

Read more on Adaptive Systems in this interview with Ramesh Rao.
[Q&A]

The team considers itself to be in "startup mode" and will probably come to number some 10 researchers. This project is led by Ramesh Rao, Calit² UCSD division director and professor of Electrical and Computer Engineering in the UCSD Jacobs School. The team includes Geof Voelker in Computer Science and Engineering, and Pam Cosman, Sujit Dey, and Rene Cruz in ECE, plus several Ph.D. students.

"Our next step is to adapt the prototype we've developed to the UCSD CyberShuttle to support wide-area coverage in a mobile environment," says Mishra.

One of the techniques being experimented with in this project is adaptive video coding. Researchers are studying tradeoffs to achieve adaptivity through various combinations of using a smaller picture, less color, different encoding (compression) methods, and so forth.

They're also experimenting with providing "hint" information to an application when a near-term need to switch network technologies is anticipated, say, for example, when the user is crossing the boundary between two technology support areas. The theory is that this hinting mechanism might enable a smoother transition for the user because the adaptation can take place over a longer period of time.

"We're also trying to accommodate the needs of applications - and the implied network load - depending on whether they require one-way or two-way communications," says Mishra. Streaming is one-way, but interactive applications, such as games, are two-way.

"We need to understand what's going on from a theoretical point of view, so simulations are very helpful. But networking technology advances only through lots of trial and error," says Johansson. "That's the sad reality. That's why we're especially looking to engage students with their creative ideas and endless enthusiasm to put their ideas to the test."

In addition to other funding, this project recently was awarded nearly $1 million from California's Industry-University Cooperative Research Program-making it the second-largest recipient of that program's UC Discovery Grants in 2002.

For more information, see http://adaptive.ucsd.edu.