Networking that goes beyond the traditional analog phone line and cable TV coax typically attached to the side of a building has been a fairly expensive option for small offices and an outright extravagance for homes. Until a year or so ago, many in the wireless industry thought that the small office-home office (SoHo) and home networking markets would be the last segments to embrace wireless technology. Conventional wisdom has held that easy-to-use products at the necessary prices wouldn't be available until well after the enterprise and vertical markets had matured. But a convergence of technology advances has led many to now predict a much earlier adoption of wireless networking in the SoHo and home segments.
Myriad advances and trends are responsible for accelerating this scenario. Among these are a rapid progress in standards development; the use of standards-based specifications, with the attendant benefits of interoperability; the trend toward larger-bandwidth pipes to homes and small offices; the development of much more highly integrated and better-performing ICs, and, finally, the entrance of major OEMs with well-exercised distribution channels in broad consumer markets. In addition, the wide availability and acceptance of cordless and mobile phones has resulted in the average consumer's becoming increasingly more comfortable with wireless technology.
An unusually long span of business growth and personal domestic prosperity has enabled consumers in small offices and homes to buy computers, mobile phones and a variety of peripherals and entertainment equipment. The advantages of connecting all this equipment are clear. Also clearly established are connectivity needs such as accessing data from the Internet, exchanging slide presentations, conducting e-commerce, checking stock prices and exchanging e-mail. Wireless services in particular have become recognized for their ability to provide connectivity regardless of geographic location and to obviate the need for remodeling buildings to install additional wiring.
Closely associated with the desire for more communications connectivity is the need for connections with greater data rates. The increasing availability of multimedia content implies that the volume of content exchanged between any two points will continue to grow.
Those local service areas within the home and office define the domain of the wireless local area network (WLAN) and wireless personal area network. Today, the most prominent methodologies within that domain are IEEE 802.11 and Bluetooth.
Wireless LANs operating to the IEEE 802.11 standard represent to many the industrial-strength version of WLANs. This group of standards provides data rates from 1 to 54 Mbits/second. Bluetooth, which is nearing commercially availability, stems from a need to wirelessly connect PCs and cell phones. A common thread in the expected success of these two solutions is a broad base of vendor support, complete and clearly defined technical specifications, good user experiences with installation and operation and attractive prices.
IEEE 802.11-based WLANs use a shared-medium, listen-before-talk protocol, collision sense multiple access/collision avoidance (CSMA/CA), which covers a large area (about 100 meters in diameter) and is designed for high-speed, best-effort packet data transfers. The WLAN's shared-access capability is based on a nonreservation protocol, which does not currently offer a complete set of Quality of Service (QoS) guarantees, although it does provide a mechanism for fair access by each node. The IEEE is working on extensions to the standard to increase the number of QoS functions.
Basically, the IEEE 802.11 specification supports two different modes: distributed coordinated function (DCF) and point coordinated function (PCF). DCF is the nonreservation CSMA/CA described above. PCF is a type of time-allocated protocol that guarantees some QoS and data integrity, through some FEC encoding, to time-bounded services such as multimedia and voice over Internet Protocol.
Under PCF and with a raw data rate of only 1 Mbit/s, IEEE 802.11 can support up to six toll-grade, full-duplex voice links with less than 20 microseconds of latency. That leaves plenty of bandwidth for some asynchronous applications or other time-bounded services when operating at 11 Mbits/s.
One of the most popular modulation techniques worldwide is frequency shift keying (FSK). Because of the high volumes generated by the combination of all those standards with similar radio architecture, the price of some components has been reduced tremendously. The Bluetooth standard capitalizes on prior FSK usage and targets low-cost applications in the 2.4-GHz band. The standard encompasses three primary usage models: wire replacement, LAN access points and ad hoc networks. Wire replacement is the main application and Bluetooth is clearly optimized for it. Its modular slot aggregation approach allows it to trade off throughput in favor of QoS for isochronous voice services.
The Bluetooth wireless system is designed to provide both reservation-based communication and best-effort traffic that allows for both voice and data transmissions for personal usage areas of fewer than 10 meters in diameter. The maximum number of simultaneous voice links is low-only three-and one piconet is limited to eight active devices.
A critical factor in the success of radio communication systems is the ability of end equipment and component suppliers to continuously innovate and consequently improve performance, while also reducing price. Although enterprise, SoHo and consumer networks are selected on the basis of solutions that provide the best value per dollar, the absolute price tolerance of consumers is much lower than that of the other two. The only high-speed standard with a low enough price for the SoHo and consumer markets is currently IEEE 802.11b, at 11 Mbits/s for a retail cost of $100 to $200 per node.
Wireless radios can be viewed as a set of simplified interconnected blocks. Obviously, for any given radio system standard chosen the quality and price of each of the major components determine the overall system quality and price: A higher dollar investment yields a higher performance radio. The relevant question at this point is when will consumers be able to buy a WLAN with some relatively good performance for $100?
The radio front end consists of the power amplifier, low-noise amplifier and frequency-conversion circuitry. Because it must receive and discriminate between very small analog signals, it must generate very low noise, be as linear as possible and be isolated from other radio system noise sources such as the digital portion of the design. Its cost is obviously a function of the radio specifications established in the standard it is being used for.
In most commercial portable transceivers, the dominant source of power consumption is the transmitter's power amplifier. Linear power amplifiers have a theoretical maximum efficiency of 50 percent, so numerous nonlinear architectures have been created to drive theoretical maximum efficiency to as close as possible to 100 percent. However, the use of those nonlinear amplifiers has been limited to standards that utilize constant envelope modulation schemes. Systems based on 2FSK and 4FSK modulation can take advantage of nonlinear power amplifier designs because their waveforms typically achieve maximum power-added efficiency when operating at a constant and maximum output power. Of course, when not operating at maximum, output power efficiency is not as good. Other modulation schemes generate waveforms with significantly greater peak-to-average ratios, forcing the amplifier to operate at suboptimal operating points a large portion of the time.
Prices are moved downward by collective experience within a competitive industry as a whole. The fundamental basis of IC learning-curve improvements can be seen in the integration of more functions into increasingly smaller and fewer pieces of silicon. With regard to a complete radio system, this is possible because of four fundamental trends.
The first is CMOS lithography reductions. The goal is to increase CMOS gate density, which directly affects the cost of digital components located in the digital baseband processor, media access control and host interface devices. The economics of CMOS wafer processing dictate that even though wafer prices increase for finer submicron lithography, reductions in transistor dimensions allow cheaper transistors. One unfortunate victim of lithography reductions is data converters. These are typically constructed in CMOS to reduce power consumption, which necessitates reducing the analog signal voltages. This loss of signal range places severe design pressure to create converters with equivalent resolution and spurious-free dynamic range.
Improvements in bipolar transistors constitute another trend. The analog front end of the radio system represents a complex design problem, with efforts continually focused on reducing the noise and increasing the fT and fmax. The analog front end must oscillate, amplify, filter, shift frequency or otherwise manipulate signals in as linear-hence distortionless-a way as possible. The matched and linear behavior of relatively small numbers of transistors becomes critically important.
Another important factor is the integration of high-Q components. The integration of passives refers to the ability of a semiconductor process to place inductors, capacitors and resistors on the same substrate with the transistors. This integration reduces the bill-of-materials cost of the subsystem and manufacturing complexity. The goal is to improve the Q factor of embedded components.
Meanwhile, the overall architectural goal is to provide acceptable signal processing with minimal complexity and, therefore, minimal components. The chosen architecture will be the result of various trade-offs decided at the overall radio design level. For instance, how many stages of downconversion should the radio use? Although zero-IF or low-IF single-conversion approaches are attractive because of the reduction in required components, compromises must be considered in areas such as receiver sensitivity and selectivity.
Most of the RF communication transceivers manufactured today utilize some variant of the conventional superheterodyne approach. The receiver is usually implemented with a collection of discrete component filters and various technologies such as gallium arsenide, silicon bipolar and CMOS. The integration of all the transceiver functions onto a single substrate is a particularly challenging process. The high Q associated with the discrete components found on a superheterodyne receiver is difficult to realize at high frequency when an integrated solution is the goal.
Also, because most single-chip radio solutions attempt to push the channel-filtering function to a lower frequency, more aggressive dynamic range requirements are imposed on the IF and baseband components. Moreover, the increased dynamic range requirements imply the use of power-hungry circuits, which runs contrary to the goals of portable electronics.