Cheaper, Smaller, Better: Squeezing Multiple Antennas into Small Spaces

Transcription

1 Cheaper, Smaller, Better: Squeezing Multiple Antennas into Small Spaces Article originally appeared as an EE Times Asia Online Exclusive in January 2004 by Gregory D. Durgin and David T. Auckland Abstract Portable wireless devices are notoriously inefficient about how they use the surrounding radio power. The major obstacle to improving efficiency in next-generation wireless is the cost, difficulty, and even misperception of using more than one antenna in a small commercial device. Engineers are discovering ways to extract both performance and cost savings from receivers with closely-spaced antennas. This new antenna technology promises big gains for manufacturing small wireless devices. Introduction It is no secret that, if they had their way with the world, engineers would place as many antennas as possible on every wireless device. More than one antenna on a receiver opens the door to a host of powerful, communications-enhancing technologies. However, the economics of mass production make it difficult to place even two antennas on the same wireless device. This is doubly difficult for small, portable devices that have limited mounting space. But recent research results have challenged conventional design rules of multi-antenna integration in small devices [1]. We show that high data link performance can be maintained, even as the cost, size, and spacing of antenna elements is shrunk to accommodate the smallest of wireless devices. This technology promises big gains for manufacturing. Our discussion is applicable to any multi-antenna technology, but we will concentrate on antenna diversity. Antenna diversity techniques as well as the closely related spacetime coding and multiple-input, multiple-output (MIMO) use multiple antennas at the receiver to provide the gain and fading immunity required by the high-rate data applications of today and tomorrow. Specifically, we are looking at compact and ultracompact diversity designs. Compact antenna diversity uses antennas separated by less the 0.5 wavelengths of radiation. Ultra-compact diversity is a compact antenna diversity design that can be fabricated and sold to a manufacturer on a single circuit board or substrate. To understand the need for compact and ultra-compact diversity, we must first discuss technical and manufacturing issues for a multi-antenna technology. Included in the

2 discussion is an example of compact diversity on a real laptop housing. Our goal is to show that integrating cheap, closely-spaced antennas onto a wireless device is both highly desirable and economically feasible. The Technical Issues The cluttered environment around us produces multipath propagation, where countless echoes of radio signals arrive from every direction at the receiver. Since these waves can constructively or destructively interfere, it is possible to get deep, random signal fades in regions of space with relatively high average radio coverage. This effect, called smallscale fading, results in wild swings in received power levels even if the receiver antenna is moved over several centimeters. You can think of space as having thousands of tiny little mines that destroy communications if the lone receiver antenna moves into them. The best way to fix this problem is to use antenna diversity, where additional antenna elements provide alternate channels to compensate for faded signals. The simplest example of this is two-element switched diversity, where the receiver switches between two different antennas to obtain a non-faded signal. If these two antenna elements are designed properly, the probability that both experience a deep fade is much lower than a conventional, lone antenna. A receiver using two-element switched diversity experiences almost 90% fewer link outages than a single-antenna receiver experiences. There are two technical problems that diminish antenna diversity gains. The first problem branch correlation occurs when diversity antennas are placed too close to one another and see similar (correlated) signals. Thus, when a single antenna experiences a deep signal fade, the other antenna is close enough to experience this same faded signal and offers no extra performance gain in the data link. For b radio links or PCS handsets, classical radio theory predicts a separation distance of at least 3-4 cm between antenna elements to avoid the correlation problem. The second problem that diminishes performance (with or without diversity) is antenna inefficiency. Inefficient antennas do not absorb radio waves into the receiver effectively. This problem occurs when individual antenna elements are poorly designed or packaged. We know how to build efficient antennas in the laboratory, but mounting these antennas onto a small form-factor is challenging. Mounted antennas do not behave anything like their free-space characteristics and wind up reflecting much of the impinging radio power. This issue becomes serious for diversity because smaller, closely-packed antennas are more difficult to make efficient. Electromagnetic theory predicts a minimum size for all antennas in order for that antenna to operate efficiently. It is up to the engineer to design diversity antennas and layouts that add system gain without compromising cost or form factor. In other words, we must balance our technical goals with the perspective of the manufacturer.

3 The Manufacturing Issues When designing a multi-antenna layout for a portable wireless device, there are five design goals. In decreasing order of importance to original equipment manufacturers (OEM), these goals are 1. Low Cost 2. Small Form Factor 3. Versatile Mounting Capability 4. High Diversity Gain 5. Single Part Number Note that diversity gain ranks number 4 on the list of OEM priorities. While link performance is highly desirable, it takes a back seat to cheap, practical implementation. Antenna diversity comes at a price. The manufacturing cost of the extra antenna and switching hardware can be difficult to justify in applications with low performance thresholds. Cellular phone handsets, for example, still use mostly single-antenna designs despite the potential for big diversity gains that were documented by researchers long before the first cellular tower was ever erected [2]. Handsets with antenna diversity increase voice quality and eliminate many of the frustrating dropped calls that we all experience. But the handset industry has not (yet) justified the cost of integrating diversity antennas into all of their cellular phones. Portable wireless data devices are a different story, however. Laptops, PDAs, and nextgeneration cellular communications have much higher data rates and place strenuous demands on their data channels. With respect to small-scale fading, some of these devices are semi-mobile, spending protracted periods in regions of deep signal fades. Multiple antennas are necessary for operating high-rate data links. Diversity antennas can also compromise the form factor of the wireless device. Multiple antennas require extra mounting space to place on the device. To avoid correlation and coupling effects, we would like to space antenna elements as far apart as possible. This may not be possible with portable wireless devices, which at best provide a few tiny, irregular spaces for antenna mounts. The irregular nature of these spaces may demand different antenna designs for each diversity element, thereby complicating the manufacturing with additional components having different part numbers. In the next section, we show how to address these manufacturing issues by using special antenna technology that allows small, efficient antennas to be brought closer together. Coupling vs. Correlation When diversity antennas are brought close together, they experience increased correlation and coupling between antenna elements. Both effects are considered undesirable, so

4 conservative engineering practice tries to avoid both. It does this by placing a forbidden zone around each element a minimum distance that cannot be exceeded by neighboring antennas. This forbidden zone is actually the superposition of two zones: one for correlation and one for coupling, as shown in Figure 1. The correlation forbidden zone is elliptical and depends in part on the multipath environment [3]. Antennas that intrude into the correlation forbidden zone will diminish the diversity gain. The coupling forbidden zone is roughly circular with a radius of one-half the wavelength of radiation. If an antenna is brought into another s forbidden zone, those antennas begin to steal received power from one another, becoming more and more inefficient. The size of the total forbidden zone varies from application to application. For b antennas, it can be as little as 6 cm still too big for many manufacturers. Figure 1: Example of the conservative approach to spacing diversity antennas. Define forbidden zones around each element that exclude both coupling and correlation effects. Then bring the diversity elements together so that they sit on the edge of each other s forbidden zones. [3]

5 In 1991, Japanese researchers Ebine, et. al. noticed some interesting behavior in diversity monopole antennas [4]. These researchers found that correlation and coupling are natural enemies to one another. As antenna coupling increases, branch correlation is suppressed. It turns out that antennas brought close together lose some efficiency, but also mutually change their radiation properties. These coupling antennas receive different combinations of multipath power; they do not experience simultaneous fades as a pair of correlated antenna elements would. This is good news for diversity designs. It means that, in many cases, it is possible for antennas to intrude upon one another s forbidden zones without much end-to-end performance loss in the data link. Although the signals on each antenna element are slightly weaker, these losses are easily overcome by the diversity gains of having two or more uncorrelated signal branches. The trade-off between coupling and correlation suppression heavily favors closer antenna elements. We will show an example design in the next section using laptop antennas. Example Using a Laptop Consider a typical laptop computer with an internal IEEE b wireless modem, operating in the 2.4 GHz radio band. This wireless modem requires two diversity antennas. The only available space for mounting them is in the recessed groove the screen hinges, towards the back edge of the laptop. W will compare the performance a standard industry design and a more aggressive, compact diversity design. The standard industry design will be two Planar Inverted-F Antennas (PIFAs) mounted on opposite ends of the hinge groove, 15.1 cm apart. The PIFAs have a total volume of 900 mm 2 each. This particular mount typically requires two different PIFAs with different part numbers. The compact diversity design uses two small, identical tabletop antennas (Etenna s EA2400 Accuwave TM ) placed only 1.8 cm apart. Each tabletop antenna has a total volume of 350 mm 2. Both antennas are aligned in the same direction (i.e. co-polar) where they are most susceptible to correlation and coupling. These configurations are illustrated in Figure 2.

6 Figure 2: Top view of two antennas mounted on a closed laptop: (left) two PIFAs (Planar Inverted F Antennas) are mounted in the hinge more than 15cm apart in a standard diversity design. (right) two small tabletop antennas are mounted with only 1.8 cm of separation. Now let s compare the performance of the two designs. To this end, we will use a combination of antenna range measurements and performance extrapolation documented in [1]. The basic steps of this analysis are listed below: 1. Mount the antennas on the laptop housing. 2. Measure the complex gain pattern of each antenna in the presence of the other (to capture coupling effects). 3. Use an indoor multipath model (the uniform hemisphere model in [1] to tabulate typical gain and correlation statistics between the two antennas. This simulates the effect of small-scale fading. 4. Use the empirical mapping in [5] to estimate the throughput and outage probabilities of a typical IEEE b link. Upon completion, we will have a semi-empirical distribution of throughput and outage for the pair of diversity antennas. The results for each case are tabulated in Figure 3. The vertical axis is the probability that the IEEE b link performs below the corresponding throughput listed on the horizontal axis (in kbps). For example, a single standard PIFA (the solid black curve with no diversity) has less than 1500 kbps about 60% of the time in a fading channel. Each graph in Figure 3 reports performance statistics for the individual antennas as well as statistics for the two antennas used with selection diversity (the dashed blue curve).

7 All curves assume the worst-case scenario: operating on the fringe of coverage in a noise-limited environment. The intercept that each curve makes with the vertical axis is the most important piece of information. This is the outage probability the likelihood that communications cannot be established. In reality a user does not perceive much drop in data connections between 1000 kbps and 2000 kbps. But an outage is sure to frustrate the user. The outage probability for selection diversity using the standard PIFAs is 1.3%. The outage probability for selection diversity using the compact tabletop antennas is 2.8%. This is still a substantial increase over the no-diversity case which has an outage probability of 10%. It is only a small, acceptable degradation compared to the standard PIFA configuration. This is far below what conventional theory predicts as the outage probability for co-polar antennas with 0.14 wavelength spacing. Furthermore, the two tabletops are smaller, cheaper, and could be manufactured on the same piece of circuit board to make an ultra-compact design. Conclusions and Implications Clearly, ultra-compact designs are feasible and economically attractive to the manufacturers of portable wireless devices. The diversity gains are possible, even in the most cramped form factors by taking advantage of coupling effects. This is quite a different mindset from current design practices, but we anticipate the day when manufacturers can cover their devices with numerous cheap antennas. Such a design would be extraordinarily effective at using the surrounding multipath power to perform communications. As it stands now, most portable wireless devices waste a lot of space. References [1] D.T. Auckland, W. Klimczak, G.D. Durgin Auckland, Maximizing Throughput with Ultra-Compact Diversity Antennas, IEEE Vehicular Technology Conference, Orlando FL, October [2] W. C. Jakes (ed). Microwave Mobile Communications. IEEE Press. New York, [3] G.D. Durgin, Space-Time Wireless Channels, Prentice-Hall, Upper-Saddle River NJ, [4] Y. Ebine, T. Takahashi, Y. Yamada. A Study of Vertical Space Diversity for a Land Mobile Radio, Electronics and Communications in Japan, vol 74, no 10, pp [5] B. Henty, Throughput measurements and empirical prediction models for IEEE b Wireless LAN installations, Virginia Tech MS Thesis, 2001.

8 Standard PIFAs Compact Tabletop Antennas Figure 3: Comparison of antenna diversity outage and throughput for the standard PIFA configuration (top) and the ultra-compact tabletop antennas (bottom). The outage probabilities for this compact diversity pair are much lower than theory predicts.