Thanks to coming advances in wireless communications, you may never have
to miss another important phone call because you were at a meeting.
Analog Devices, Inc., Norwood, MA is developing wireless local loop (WLL) technology that would replace the wired PBX phone in your office. The system would include a cellular base station in your office building or manufacturing facility, and an antenna that would be able to pick up and broadcast conversations.
"Essentially," says Rupert Baines, a communications engineer and marketing manager, "you could take your office phone to the meeting with you, or to the cafeteria, or anywhere else in the building."
The technology is based on the DECT (Digital European Cordless Telephone) standard, already used by about 100,000 subscribers overseas, but not yet in the U.S.
Panasonic, Inc. has also developed technology for turning office phones into cellular phones, and vice versa. Its BusinessLink Personal Communications System includes a switch that goes behind the PBX system. "We re-direct the circuit to our switch and associate it with a wireless handset," says engineer John Avery. The result: You can take your office phone out of the building and it converts to a cellular phone. Come back to the building and it reverts to a regular office phone again.
These are just two of the many breakthroughs in wireless communications that have spawned the era of the "virtual" office--making your car and your home as efficient places to conduct business as the traditional four-walled-and-carpeted office.
Offering faster processing speeds, better quality picture and sound, a slew of new features, and weight and cost savings, the next generation of wireless products will go far beyond the capabilities of today's systems. "We've barely scratched the surface," predicts Samuel May, a wireless communications specialist with the Yankee Group, Boston. "The big movement involves designing systems using innovative engineering that may someday make today's standards--CDA, TDMS, or AMPS--transparent to users. You will be able to use wireless anywhere, the standards simply won't matter."
Innovation lies behind the products that will pave the way for next-generation wireless, personal communications systems (PCS). Let's take a closer look at some of the engineering feats that will make this happen.
Radio functions in software. One group of communications engineers are betting on a new branch of radio services that don't depend on hardwiring. Called software radio, it is based on advances in multi-band antenna, radio frequency (RF) converters, digital signal processors (DSPs), and the use of intermediate frequency (IF). Aimed at cellular networks, software-radio architecture offers several advantages over wired radio.
For instance, many users can share one radio channel, and, with multichannel users, comes considerable cost savings. Programmability also leaves room for more flexibility between different standards, as well as offering designers different ways to optimize a wireless network.
"We expect software-radio architecture will lead designers to higher levels of complexity needed for the next generation of wireless communications, such as PCS," says Joe Mitola, a consulting scientist with Mitre Corp., McLean, VA. "It will be designed into base stations for integrating the modes needed to make multimedia happen." Using voice, fax, data, or multimedia, a mobile software-radio terminal will directly interface to the user, while the base station will interface to the public switching telephone network.
Despite its name, a software radio actually uses a lot of hardware. The building blocks for a typical system include: a power supply, an antenna, a multiband RF converter, a chip with on-board A/D/A/ converters, a general-purpose processor, and the memory for performing the radio functions. So, why call it "software" radio? "Everything is programmable," explains Analog Devices' Baines. "That includes RF bands, channel access modes, and channel modulation and demodulation schemes. The key is using programmable components, such as high-speed DSPs and ADCs."
Consisting of printed circuit boards, Analog Device's evaluation system is up and running. Measuring 8.5 x 11.5 inches, it contains a receiver converter, a digital down converter, and off-the-shelf A/D and DSP evaluation boards. Cutting out a lot of the hardwiring also means that the A/D/A converters can be placed closer to the antenna, thus significantly reducing the potential for interference problems.
"Several process stages previously used to stabilize, amplify, and filter the signal are eliminated,"Baines adds. "While you gain more flexibility by cutting out all of the hardwired stuff that came between the antenna and the A/D/A converter, the closer the converters are to the radio, the less problems you're going to have with noise and distortion."
A key component in the software radio prototype is Analog Device's new 12-bit ADC, the AD9042. With a sampling rate of 41 Msps, it boasts speeds 60% faster than current chips on the market. It also requires one-third less energy, taking 575 mW off a single +5V power supply. "The chip is designed for direct intermediate-frequency sampling and oversampling of wideband signals," explains Baines. "This makes it an excellent candidate for the tough requirements of software radios, such as the next-generation, wireless base stations that can understand multiple standards (AMPS, CDMA, or TDMA) merely by changing a program in the DSP."
The converter uses a two-stage architecture with six- and seven-bit converters. It is based on a cascaded-magnitude amplifier design that does not require any laser trimming or calibration to achieve the 12-bit accuracy. Recent successful tests included a broadband signal containing up to 48 different tones.
Leaping hurdles. Still, despite its advantages, many challenges stand in the way of commercialized software radio. "The ADC and DSP cores needed for this kind of architecture consume a lot of power," cautions Mitre's Mitola. "Design engineers must think of more innovative ways to keep these chips cool. In addition, these systems will need ten times more ASIC processing power than hardwired systems." Other design obstacles: engineering the needed wideband low-loss antennas, and complexities resulting from mixing and matching software tools.
Even considering such design limitations, the consensus is that this technology may prove vital in future wireless links. While it currently revolves around base stations, some experts believe software radio will eventually find its way into the handheld set, with low cost as a very tempting goal. "I do expect we will see small, cheap software radios that fit in your palm," predicts Joseph Kennedy, director in charge of software radio engineering at Engineering Research Associates, Vienna, VA. "You'll be able to connect with any air standard out there."
An antenna with smarts. The first application for software-radio technology is on the market in the form of a "smart" antenna for the 800 MHz commercial mobile cellular radio band. And Kennedy and his colleagues at Engineering Research Associates will introduce a highly accurate signal-location system based on a self-adapting antenna. "This type of antenna technology simply could not be realized in a hardwired radio," says Kennedy, who heads the CAPITAL project.
Basically, the system tracks and analyzes traffic flow by monitoring cell phone signals from cars. Since cellular phones are popular consumer products, their signals represent a fairly good indication of traffic flow. "Looking at the technology options, we decided that it would be more cost effective to devise a signal-location system geared to mobile cell phones, rather than install roadside traffic monitoring systems," Kennedy explains.
The CAPITAL project involves a mainframe software system that resides at several locations in the Washington, DC, area. A given cell-phone signal is pinpointed by comparing its time of arrival at different locations. The system also uses Global Positioning System (GPS) data, as well as advanced signal-processing functions made possible by having the radio implemented as a software algorithm.
Rather than simply receiving a signal, the antenna function in the software radio remains active, scanning frequency bands and analyzing the nature and power of a complex welter of signals. Reception and discrimination of a signal is enhanced by generating a "null" function at other angles, which screen out noise.
These sophisticated signal-processing functions must sort out a given cell phone from the massive flow of radio signals. Combining all of the signal-processing location methods allows CAPITAL to pinpoint a single cell-phone signal to within 10 meters. At that resolution, individual vehicles and their movements can be resolved and traffic flow statistically analyzed.
The design is targeted at cell phone base stations to increase the number of available channels. To make this happen, Kennedy points out key differences between the smart antenna and conventional phase-antenna arrays. "Conventional antennas use steer beams or null, and designers are primarily interested in increasing power," he notes.
Rather than concentrating on power, Kennedy and his project team are focusing on maximizing the Carrier-to-Interference (C/I) ratio. "It's the interference from adjacent cells that hurts the performance of today's cellular phones," he explains. "If you increase the C/I, you will have more channels available in a geographical area. We've also included multi-path conversion, which plays a big part in cellular evolution."
Field tests are currently under way, including police and paramedic services to locate someone in distress. Anyone experiencing difficulties could simply leave their phone on after making an emergency call; the antenna would inform an emergency vehicle of the exact location.
With the complexity of wireless networks comes a load of problems stemming from new standards and protocols needed to keep up with the exploding market. Even for expert network designers, understanding the performance and behavior of a particular network, and then building the infrastructure to get a project up and running, can be extremely time-consuming.
Designing the networks. "It takes months just to write the Fortran code for the infrastructure alone," says Marc Cohen, co-founder of Mil 3, Inc., Washington, DC. His firm offers designers a sophisticated network toolset to simulate and analyze the performance of wired and wireless communications systems.
Based on a three-level hierarchical structure, Mil 3's product, OPNET, makes it easier to analyze and simulate complex networks. It resulted from a thesis project spawned by Cohen's brother, Allain, and a fellow student at MIT. "After graduation, we decided to commercialize it," says Cohen. The result: a software simulator that features over 300 functions. The client list includes more than 700 companies that range from small office-equipment design to fast packet-switch satellites.
"Each environment has its own set of problems," says Cohen. "Even in a small office you are up against a large amount of signal interference and complex propagation issues. If you're are in the city, high-density buildings create added multipath and reflection problems."
Running on X Windows, and supported by SUN, DEC, IBM, and Silicon Graphics workstations, OPNET presents the user with icons and menus that combine detailed network analysis with radio-frequency techniques. "In the past, these issues have been treated separately," Cohen relates. "By combining them in the same package, you get a more powerful analysis of the performance of your network."
The highest hierarchical level, the Network Editor, graphically captures the physical topology of a network. "You can create nodes that represent the different communicating sites in the network," Cohen explains. "each node fits into a class that defines its attributes, such as a switch's processing speed, a workstation's traffic-generation rate, or a gateway's buffer capacity."
The nodes communicate using several kinds of communications links. For wireless modeling, the company offers an OPNET Modeler/Radio link. It includes a feature developed by NASA that calculates satellite orbits specified by their six-element ephemeris data sets. It also takes into consideration the effects of solar and lunar gravity, atmospheric drag, and solar radiation pressure.
On the second level, the Node Editor graphically captures node architectures as diagrams of data flow between modules that represent hardware and software subsystems. The modules are divided into two categories. For dedicated functionality, traffic generators, transmitters, and receivers are configured on the menu. In the user-programmable behavior module, the functions performed by queues and processors are user-specified.
The third tier consists of a Process Editor. It's based on a state-transition diagram for supporting different protocols, applications, algorithms, or queuing policies. The states and transitions graphically define the progression of a process in response to events. For each state, a library of about 300 functions exists to implement the general logic for your applications. Simulation results can be plotted as time series, scatter plots, histograms, or probability functions.
Booming applications. The explosion of applications and customers in need of detailed network analysis has even stunned Cohen. "We were amazed at the variety of uses engineers are finding for the tool," he adds.
Network simulation is the key engineering tool for engineers at COMSAT Laboratories, McLean, VA, a pioneer in the development of future communications satellite technology. "With the rapid emergence of point-to-multipoint services, such as video conferencing, there is a growing need for on-board fast packet switches," says D.J. Shyy, a research engineer at COMSAT. "The programmability allows us to develop accurate models of multicast fast packet switches." The switches were examined under different traffic patterns and loading scenarios. The engineer analyzed queuing behavior, call-splitting capabilities, and congestion control.
Hughes Aircraft Co. also uses the system to develop intelligent communications controllers for multimedia tactical networks. The controllers integrate various satellite and HF communications systems under a common control concept for providing a flexible network service. OPNET not only models a variety of radio and network characteristics, but also functions as a development tool for constructing the advanced protocols for Hughes' customers.
Meanwhile, engineers at Adroit Systems, Inc., Alexandria, VA, use OPNET to model linked air-ground networks. These networks involve disparate heterogeneous systems that range from local area networks (LANS) and wide area networks (WANS) to high-rate TDM RF links.
It's all in the chips. To compete in such a highly competitive market, electronics companies must load more horsepower and memory on a chip. "Designing for the wireless market is different ball game," says Thomas Brooks, DSP marketing manager at Texas Instruments, Dallas. "DSP cores that are optimized for wireless communications bring end-product advantages, such as higher performance and lower power dissipation per function to the OEM."
Equally important, the push for digital technology remains strong. "You get more calls per channel, more capabilities, and privacy and safety features that are better controlled in a digital environment," Brooks adds. "Plus, with performances of up to 50 MIPS, the OEM has the ability to incorporate such value-added features as a speaker phone, noise cancellation, voice dialing, and short messaging."
Texas Instruments currently has under design chips with larger memories, core architectures, and intelligent peripherals. "The larger on-chip memory means having the ability to adapt to new industry standards and using fewer components per system," says Brooks. TI's designers also use thin packaging, glueless interfaces between the DSPs, analog components, and the speaker/microphone for designing smaller, lighter equipment at a lower cost.
Motorola also will come out with more wireless-specific DSPs. The latest line includes the DSP56300. It features speeds of 66/80 MIPS at 3.3V. "We're designing for the future with a component like this," says Mark Reinhard, vice president and director of technology for Motorola's Semiconductor Products Inc., Schaumburg, IL. "The migration path will be up to 100 MIPS and 1.8V within two years."
While digital technology, with its flexibility and programmability benefits, will reign as the key processing technology, demands for mixed-mode processing have many companies keeping analog on the market. "We expect analog to be around for at least another 20 years," says Russell McDonald, manager of TI's mixed-signal sector. Currently under development is a new power-saving chip, the Advanced RF Circuit Telephone Interface Circuit or ARTIC-136. It will hit the market in early 1996. A GMS Baseband Interface Circuit is planned for later in the year.
Analog Devices also hopes to attract the mixed-mode market with its AD7015. The company claims that the device is the first complete 3V codec for Global System Mobile Communications (GSM) handsets. It integrates all of the mixed-signal components: the voiceband codec, which connects the microphone and ear piece to the DSP, and the baseband codec, which connects the DSP to the radio, filters, amplifiers, and auxiliary converters. Previous designs have required several devices and 5V supplies.
Such developments, backed by innovative engineering, will play a major role in building the infrastructure and dynamics for the next generation of wireless technology. The results will propel wireless into the next century.