New Tech Press recently interviewed Fanny Mlinarsky, President of octoScope (www.octoscope.com/),the manufacturer of octoBox small anechoic test stations and octoFade channel emulation solutions, to talk about the latest in radio baseband technologies.
NTP: There are already some complaints internationally from consumers about the LTE standards some device manufacturers, like Apple, are choosing. For example, the new iPad works on the Verizon and AT&T networks, but not on T-Mobile. Even Australia has significant limitations. How do manufacturers choose networks?
Fanny Mlinarsky: Decisions, decisions… Modern MIMO (multiple input multiple output) radios vastly enhanced range and throughput performance results from their ability to adapt modulation in real-time. New generation 802.11n/ac and LTE/LTE-Advanced radios make decisions on a packet-by-packet basis, selecting the most appropriate MIMO techniques for the channel conditions. The baseband layer in the radio is responsible for such decisions.
The baseband DSP (digital signal processing) algorithms have to keep pace with the fast-changing airlink conditions. MIMO techniques include spatial multiplexing to increase throughput in good conditions, transmit and receive diversity to improve link robustness in adverse conditions and beamforming to enable multi-user MIMO or to increase operating range.
NTP: Let’s break this down. What's the benefit of spatial multiplexing?
Fanny Mlinarsky: Spatial multiplexing has the potential of multiplying data rate at the PHY layer by enabling MIMO radios to send multiple data streams simultaneously in the same space and on the same RF channel. Spatial multiplexing typically requires channel conditions that are characterized by low correlation (explained below) and high SNR (signal to noise ratio). Theoretically a MIMO link with two or more radios in both the transmitter and the receiver can support up to double the data rate of a SISO link if conditions allow. Three or more MIMO radios can theoretically triple the data rate; four or more radios can quadruple the data rate, and so forth. Due to overhead in the protocol, throughput at layer 3 is always lower than the data rate. In practice, due to non-ideal conditions, throughput is typically below the multiple of the spatial streams. Today’s 802.11n technology is limited to three streams for a 3×3 MIMO implementation.
Correlation refers to the ability of a MIMO system (radios and RF channel) to support multiple streams. The lower the correlation, the more likely it is that the channel can support multiple streams. Usually the further apart the antennas are spaced from one another in a MIMO device, the less correlation exists in the MIMO system. Cross-polarizing MIMO antennas (i.e., having one antenna vertically polarized and the other horizontally polarized) also lowers the correlation. Multipath in the radio channel also serves to lower correlation as do transmit diversity techniques.
NTP: What's the application for transmit and receive diversity?
Fanny Mlinarsky: Transmit diversity techniques are used in poor channel conditions (e.g., high SNR) to improve the robustness of the radio link by minimizing packet error rate and thereby increasing throughput and operating range. Transmit diversity algorithms generate time- or frequency-orthogonal versions of the same data stream with the goal of optimizing reception. Typically if reception of a data stream is poor, for example due to a null in the channel response, the orthogonal version of the same stream is likely to exhibit better reception. Examples of transmit diversity include cyclic delay diversity (CDD), space time block coding (STBC) and space frequency block coding (SFBC). Transmit diversity techniques spread the signal so as to create artificial multipath in order to de-correlate signals from different antennas with the goal of delivering a peak on one of the redundant streams while there may be a null on another.
Receive diversity, such as maximal ratio combining (MRC), combines signals from multiple receivers in order to optimize the chances of receiving packets error-free. A receive diversity scheme, such as MRC relies on low correlation that enables the transport of multiple distinct streams.
In a cell phone, where antennas are physically close to one another and difficult to de-correlate because they essentially see the same signal, MIMO gains in throughput are typically lower than in larger devices where antennas can be spaced apart more easily.
In summary, MIMO techniques, such as spatial multiplexing and diversity, require low correlation in order to increase throughput.
NTP: How does beamforming fit into the mix?
Fanny Mlinarsky: Beamforming enables multi-user MIMO (MU-MIMO) in the same frequency channel by virtue of shaping the antenna radiation pattern. Beamforming can also be used to improve operating range vs. omnidirectional antenna transmissions by focusing all the power in one direction. Beamforming requires an antenna array to form each beam. Typically 3 or more antennas locked and offset in phase are necessary for effective beamforming.
Typical radio communications stack. The baseband layer, responsible for DSP algorithms, is typically implemented as a combination of computational logic, firmware and software.
And finally, to make decisions more interesting, multiple MIMO techniques, including spatial multiplexing, transmit diversity, receive diversity or beamforming can be used in conjunction with one another. Thus, the choices for MIMO signaling are many and baseband DSP has a lot of work to do.
Beamforming techniques use antenna arrays to direct most of the energy at the receiver or to form multiple beams enabling multi-user MIMO.
NTP: Any major drawbacks?
Fanny Mlinarsky: Clearly, real-time decisions on which MIMO technique or combination of techniques is most appropriate for the constantly changing channel conditions rely on sophisticated baseband DSP algorithms. As with any computationally intensive logic, the drawbacks include increased power consumption and increased cost of silicon.
NTP: Adaptability of DSP algorithms means you can do more with them, but does that also increase complexity?
Fanny Mlinarsky: There is no free lunch. Adaptability increases complexity. And complexity mandates thorough and lengthy testing that has to be performed over a wide variety of changing channel conditions.
NTP: What new design considerations do these adaptable algorithms introduce for Wireless System Designers?
Fanny Mlinarsky: The implementation of adaptable DSP algorithms mandates testing that is done using channel emulators, such as octoScope’s octoFade. Channel emulators typically implement standards based channel models defined by standards committees, such as IEEE 802.11 and 3GPP. Channel models emulate multipath and Doppler fading for a variety of airlink conditions, including indoor channels (e.g., typical home and office spaces), outdoor channels, a range of velocities of the radios and reflectors (e.g., cars on a road) and other conditions. Channel emulation will be covered in the next part of this series.
NTP: How do the advanced solutions benefit engineers or humanity at large? What are we destined to do now, that we couldn't do before?
Fanny Mlinarsky: The advances in baseband DSP solutions enable us to use limited spectrum with utmost efficiency, helping high density, high-speed wireless networks deliver pervasive connectivity with performance comparable to that of wired networks. In the March 9th Wall Street Journal article, physicist Michio Kaku predicts that “Instead of one chip inside a desktop, we'll have millions of chips in all our possessions: furniture, cars, appliances, clothes.” All these devices will generate a buzz of simultaneous wireless transmissions that will ride on top of sophisticated adaptable baseband DSP.
NTP: What is the “next step” to get us to this “network of things” you mention above. Obviously spectrum space is limited, so how will we connect everything?
Fanny Mlinarsky: The challenge becomes managing traffic types and priorities in the most optimum manner. Some traffic is mission critical and real-time in nature, for example voice or collision avoidance messages among vehicles on the road. Some traffic can be transported in the background, for example email or inventory queries directed to vending machines. This is where intelligent network with the capability of prioritizing and scheduling traffic will become important. For example, a service provider may set the cost of video downloading to be much cheaper at night than during prime time to discourage unnecessary traffic loads. And for handling congestion, IMS (IP multimedia subsystem), which is soon to be deployed by Verizon, can provide many levels of QoS and sophisticated prioritization of traffic.
NTP: The recent release of the latest iPad has everyone buzzing about the speed of 4G, but more importantly, it’s increased the discussion of what 5G might look like. Can you describe for us what you think 5G might look like and how you expect us to get there?
Fanny Mlinarsky: Some vendors of the emerging IEEE 802.11ac devices are already touting their new generation Wi-Fi technology as 5G. This technology is characterized by sophisticated spatial multiplexing in conjunction with beamforming and wider channels of 80 MHz being demonstrated now with 160 MHz possibly a few years hence. Currently the widest channel width used by deployed 802.11n products is 40 MHz. And since Wi-Fi is now also used outdoors as wireless backhaul for small base stations, 802.11ac with its beamforming capabilities and up to 8×8 MIMO may be the solution for indoors and outdoors. For backhauling, beamforming can be used to extend the range between base stations. ■
Learn more about Footwasher Media