Q&A with head of Rutgers University's wireless lab on topics ranging from municipal Wi-Fi's challenges to the overhyping of RFID.
Dipankar "Ray" Raychaudhuri, a professor at Rutgers University and director of its Wireless Information Network Laboratory, describes himself as an "average" wireless user. His knowledge about wireless, however, is anything but average, which explains his involvement in such efforts as the National Science Foundation’s GENI project. Raychaudhuri recently talked with Network World executive news editor Bob Brown on topics ranging from municipal Wi-Fi's challenges to the overhyping of RFID (for more on WINLAB, read our companion story.
Where is wireless heading?
Wireless technology is becoming a lot more heterogeneous in that there are a lot of standards that are beginning to coexist and each is finding its sweet spot. For a while it looked like there was only cellular, then Wi-Fi came into the mix, then other technologies like WiMAX began to look viable. [Read Network World's guide to next-generation wireless here.] We're entering a period where multi-radios are going to be the norm. Enterprises are mostly using Wi-Fi now internally, but even there, there are so many standard evolutions and a couple of revolutionary new things in the next few years which should change many of these things. Broadly, our thinking is that we need to deal with a much more complex wireless world where you have different, physical-layer wireless-radio standards and a variety of networking modes including multi-hop and ad hoc types of intelligence that aren't used very much today.This gets into your work on pervasive wireless [in which different wireless technologies interact automatically]?
The vision I've outlined is still a long-term road map that doesn't happen in a year or two, but we've seen a lot of progress in a couple of areas. The pervasive computing, of course, is not so much an enterprise application yet, but there are a lot of vertical industries where pervasive computing is coming in different ways. In supply chain management and RFID, it is maturing quite rapidly. We think that will make it possible to do more advanced inventory control. For example, one of our guys is working with the jewelry industry, where the classic, passive RFID doesn't work that well but they want to do things, like know when an object has been removed and have a continuous reporting of presence that comes out of active RFID devices.
Some people are really down on RFID these days. You're not?
It got hyped too much, and like most things in our industry, it takes a couple of cycles of hype and gloom. Things are improving and I'm seeing the trend that now at least for high-value applications -- like this jewelry one where you have a relatively high loss rate -- they have quite an incentive to use new technology like this. The other area where we're seeing a lot of traction is in the vehicular industry with companies like GM and Toyota looking for wireless to hook into their systems for things like automobile safety and collision avoidance and some car-to-car convenience applications, such as sharing media and getting navigational updates.
Another wireless application that's taking a beating these days is municipal Wi-Fi . . .
The muni Wi-Fi business case has not been made yet. The mesh products first released still suffer from technical issues, many of which can be improved. The problem is that most of the mesh believers have offered a speed that is well below what people consider broadband. They've been typically sold on the notion of it's going to be a low-cost or free service so people wouldn't care about the actual speed. But my experience is that no matter if it's free or not, people expect state-of-the-art performance. A lot of the mesh companies are working to tighten up the products, including their performance. The problem with mesh as deployed is that the technology becomes interference-limited rather than bandwidth-limited. [Learn more about wireless mesh products in Network World's Wireless Mesh Buyer's Guide.]
So has the problem been more of a technical or business one?
It's a combination of the two. The value proposition hasn't quite made it in the United States so far. But this kind of technology is doing quite well outside the United States, especially in less-developed areas where there is less wiring available.
As for the business case, this notion of municipalities running public services in the United States has never really worked. Even for fiber -- this was tried by quite a few municipalities, and most of them backed out from that during the broadband days. Right now in the United States, the broadband availability footprint is quite good, so the number of areas where no broadband is available and muni would be the right solution is decreasing.
A lot of these early deployments are limited by the speed, too, so if you want to change the business model from free to paid, you have to pay more attention to performance. And that's something at WINLAB: We've spent quite a few years looking at: fundamental issues of how to improve multi-hop wireless performance. [Read about muni Wi-Fi vs. 3G and the rest of networking’s 50 greatest arguments.]
While RFID and muni Wi-Fi have taken their lumps, 802.11n seems to be getting a more positive reception. What are you seeing on that front?
We haven't yet been able to do a comprehensive test of "n," but we have some work going on right now. I think you'll see a stepped change in improvement with "n" in a lot of office environments even though the upper-end claims on bit rate aren't very easy to achieve. But simply having parallel channels gives you a lot higher speed. So we think this kind of technology on the one hand is good and it's going to work. On the other hand, the 802.11 standard is sort of reaching a point where completely new solutions are being discussed also. [Read about the first large-scale 802.11n wireless LAN rollout.]
What's your assessment of where wireless security stands?
With wireless LANs we've reached a point where there are security products available but you still have lots of people putting their wireless deployments outside the firewall. Wireless, of course, makes it easier to attack, and there are lots of honeypot access points and rogue [access points] out there. What's interesting about wireless is that the radio medium carries a lot more information, which in a sense makes it easier to detect and prevent intrusions. The bad news is that the medium is more open. . . . We don't think the security problems are solved, and we have several projects going on at WINLAB. [Professor Wade Trappe has been doing a lot of work on wireless security.] We're focusing on problems in which you can use knowledge of the radio-level properties to detect an attacker or prevent intrusion.
How high are you on femtocells [essentially, small cellular access points that route nearby wireless voice traffic through preexisting broadband connections]?
Femtocells are reliable, and they work. The main problem is the frequency; there are not that many channels. If you do a small cell site in your home or office, you have to make sure that signal doesn't interfere with the neighboring home or office or with the wide-area transmission. So that problem is not solved; especially for [Code Division Multiple Access] technology; there's not enough bandwidth to go around unless you can come up with some clever idea to detect signals.
That kind of brings me to the spectrum area that we have a lot of work at WINLAB on: how to dynamically share spectrum coordination techniques, for example.
A recent Science Board Report indicated that the United States has work to do in terms of funding basic tech research. Is the U.S. putting enough funding behind wireless research?
Typically, if you ask a professor about funding, they will say it's not enough. If you compare the U.S. to Europe, the amount of funding here is smaller for wireless; they've had more of an applied funding approach, and they've done well with that formula. They have at least $150 million to $200 million in basic research funding in Europe per year and in the U.S., the NSF might be funding a total of $20 million to $30 million earmarked solely for basic wireless research. Other agencies might account for another $20 million or so. Total basic research funding in the U.S. is not large, but on the whole I think it's working OK. We have support from some companies also. In terms of growth areas, there are initiatives called GENI [Global Environment for Network Innovations] and FIND [Future Internet Design] at the NSF right now on the future Internet, and there's been some money put into this area, and wireless is a big part of the future Internet.
What most excites you and most frustrates you about wireless?
For excitement, we are working on something called cognitive radio and soon we'll be starting to see some light at the end of the tunnel. This kind of work on fully programmable software radios has been going on for a long time, but a lot of factors are now coming together -- the complexity of chips has come to a point where it's really possible to achieve a cognitive radio functionality. We're now working on this fully software-defined radio which has a lot of nice features with immediate applications to markets like vehicular communications. We are building prototypes. We're also upgrading our ORBIT [Open Access Radio Grid Testbed] to include these second-generation, software-defined radios by the end of the year so people can try them out.
What frustrates me about wireless is there is often too much buzzword-driven marketing of particular standards. Deeper, technical discussions sort of get pushed away because people spend too much time arguing about if WiMAX is better than LTE [Long-Term Evolution]. I think we'll come to a point eventually where these things will settle into a set of good technologies that are more programmable than they are right now. In another 10 years we'll see a simpler landscape.