NF: Softwareespecially EDA softwareis increasingly playing a bigger role in test and measurement. How do you see that trend growing?
JM: There's currently a disaggregation of the design process as manufacturers focus on being a system integrator and trying to build software ecosystems. For example, handset designers have sockets and they want vendors to compete for those sockets. They need a power amplifier (PA) that will work. They don't have time to write a detailed spec. Instead, they tell their vendor, "I want a PA with a bit error rate of X with these wireless standards." The people designing the ICs want to win the socket and be the first to prove that what they're doing is going to work. That's driving an interesting ecosystem change. We have found that there's a big inefficiency around the communication between the IC vendors and what we call system integrators. Right now, they send these gigantic specifications back and forth and they'll specify, "Do a part like this." Then the vendor takes it and tries to understand it. The vendor designs it, fabricates it, and tests it like crazy. That testing can take a big lab of test equipment, lots of time and people, and it generates a giant report.
NF: Did you find a better way to communicate what the system integrator wants versus what the IC can deliver?
JM: We've brought in nonlinear X-parameters so the system integrator can use some EDA tools to specify what kind of behavior they would like from that part and send it to the chip designer. If this ecosystem makes it easier to communicate back and forth, the system integrator could work with many more vendors. The first vendor back that can prove that their design works will win the socket. To do this quickly, the vendors can develop their PAs in our EDA software, which will enable them to generate an X-parameter model of the schematic. Rather than going through a fab cycle, they can just push a button and say, "Here's how it will behave" and send it back up to the integrator, who can run it in their simulation and say, "yes, but I want this" or "something has changed; now I want something different."
NF: It's been six months or so since Agilent announced its X-parameter capability. What has been the market feedback?
JM: A lot of customers have said, "Prove it. Here are some parts, show me the model, and I'll take it back to my lab" and they basically run that cycle back and forth. But it's becoming pretty popular with the customers that have it and we're starting to see sales of the nonlinear vector network analyzer (NVNA). The ADS version that has the X-parameter generator came out in the fall and customers are excited about it. The other feedback we're getting is from IC vendors, who really like the fact that intellectual property does not change hands. If you have a design and you generate this model and send the model across, they can't reverse-engineer the design.
NF: With the advent of Long Term Evolution (LTE), I'm sure you have seen an increased need for both test and software.
JM: With an LTE implementation, you can have a bunch of radios that are all relatively the same frequencies and jammed in a very, very small space and they're all transmitting and receiving. How do you keep one from blowing out the other? And you have to interface it to an antenna or a bunch of antennas, which brings in all of the multiple-input multiple-output (MIMO) challenges. People don't really know what's going to happen. It is the ultimate EDA opportunity. We're making big investments in 3D electromagnetic (EM) simulation because that's a foundational technology for the people doing this kind of work.
NF: Could you tell me a little bit about the EEsof group versus the test and measurement groups in Agilent and how much you coordinate and work together?
JM: I like to tell customers that we eat what we cook. We're unique in the EDA world in that we have customers right down the halla lot of them. My office is actually right over a gallium-arsenide (GaAs) foundry. So we get to see the trends in the high-frequency and microwave areas. But we run the EEsof group just like a business. I use a lot of synergies with the instrumentation. My software will run inside some of the latest instruments. Let's say a customer wants to use LTE but try a different version of encoding. They can use my software to modify the LTE standard. This software will connect into Agilent instruments and generate it. Then you can take the instrument and hook it up to an antenna and now you've got an extremely expensive but very accurate transmitter that you can play around with and see how it works. Some of the really deep R&D folks in leading companies and universities really utilize that capability because it gives them a lot of flexibility to develop new standards.
NF: When you talk about the changing wireless standards, it brings to mind the increased use of software-defined radio (SDR) in test equipment.
JM: We've been seeing the same thing with the field-programmable gate arrays (FPGAs) and the enormous capacities that have been coming online. Plus, designers are pushing to do more in digital and digital signal processing (DSP) in that whole transmit and receive area. That opens up the flexibility for SDRto change that radio on the flyespecially for people investing in base-station infrastructure. There's also the homeland security/defense need. They want to be able to quickly set up or monitor communications in different areas. If you look at the two wars that we're in, remotely detonated roadside bombs are also a huge issue. Part of the SDR is having a product or device that can quickly see what's out there and then configure itself to be able to receive or interrupt.
Continue to page 2
NF: Everyone looks at Agilent as a leader in both test equipment and EDA. To what do you attribute all of that constant innovation?
JM: It comes from the Hewlett-Packard roots. We hire bright people and we appreciate bright people and new ideas. It's a very open environment. Everyone has open-office cubes. How do you create an environment where you can innovate? You end up building an organization that is really good at particular areas and you findespecially with social networking these dayspeople who like work of a certain type. Let's say they really like simulation. They'll figure out that Agilent has some of the best simulators in the world and say, "I want to work there." Then we connect and we bring them in.
NF: Aside from bringing in great talent in terms of the individual engineer, Agilent has made some great acquisitions. Can you talk about some of those and what they've helped you bring to the table?
JM: We did two acquisitions in the last couple of years: Eagleware and Xpedion. The attrition from those is extremely small. It's one or two people. Especially with software, you're not buying a special machine that makes something. You're buying people who have great ideas and source code. You have to keep the people, leverage their ideas, and move people into different positions so they can contribute to new areas. The Chief Technical Officer for Eagleware became the person in charge for the whole ADS platform.
With Xpedion, they had some fantastic mathematical simulation. When we bought them, we brought our tech guys together and found out that their technology was very complementary. So we took the best of both and each saw major performance enhancements.
NF: What emerging applications are you starting to hear more about?
JM: ZigBee is something we're starting to hear more about. Everyone is trying to figure out a way to connect more devices to the Internet. Another trend is wireless HDMI. A lot of companies are trying to send these high-speed digital signals to displays. This involves frequencies around 60 GHz. One of the challenges is that as frequencies go up, the width of the signal starts to beam. They're struggling with finding ways to steer that beam. Say you've got a person who's got a DVR or BluRay player and they're trying to send their signal to a flat-panel display. The consumers are going to put that display in different places and maybe move it around. So how are you going to steer that beam? We're also seeing high-speed digital designers who are encountering microwave problems, but are not microwave engineers. One example is SATA disk drives, where the serial bit streams run at wickedly fast speeds but use cheap connectors inside a PC. It's a microwave problem and they must find a way to get the signal through the low-speed cable and connectors.