By Russell J. Hoppenstein, applications manager, wireless infrastructure, Texas Instruments
In the second installment of this series, “Expandable capacity in MIMO systems,” I covered multiple-input, multiple-output (MIMO) capacity calculations. The channel solution shows that a MIMO system thrives in a multipath environment to achieve the proper spatial multiplexing. You can use this technique in advanced systems to increase capacity not just by pushing more data through the pipe but by connecting with multiple users. The expandable concept known as massive MIMO extends to a very large antenna count to maximize data throughput.
MU-MIMO
The previous installments discussed a MIMO system with an M × N antenna matrix. For mobile users, it is impractical to use multiple spread-out antennas within one device because of the physical size constraints. Recall from the previous installment that without sufficient separation, it is difficult to calculate the channel matrix to recover the signal. In applications like wireless infrastructures, the base station communicates with multiple users.
Multi-user MIMO (MU-MIMO) incorporates multiple transmit and receive antennas at the base station (where size and complexity are possible) to communicate to multiple independent users (mobile users). MU-MIMO is a form of spatial division multiple access, which facilitates adding capacity in the form of multiple users without increasing signal bandwidth. Users are differentiated by unique space-time codes.1
Equation 1 shows the space-time code matrix:
The columns in the matrix represent the space-time code of length t across k number of elements.
The system transfer function is expanded to incorporate the space-time coding matrix, expressed by Equation 2:
In MU-MIMO systems, the base station handles the processing heavy lifting, but you still have to calculate the channel characteristics denoted by the H matrix. For the uplink portion (mobile transmitting, base station receiving), the calculations to determine the channel characterization follow the inverse matrix calculation when the mobile devices provide appropriate pilot tones. For the downlink portion (base station transmitting, mobile receiving), the situation is more challenging. The transmitter must have knowledge of the channel and pre-code the signal to compensate accordingly.
The receivers get copies of all information from all users, but the receivers know their respective orthogonal codes. When the codes are applied to demodulate the signal, the information is received, and the other users’ data will spread and look like low-level interference. This process is similar to code division multiple access, wherein codes spread the signal; however, in MIMO, the bandwidth does not increase. The spreading in this case is over space (multiple antennas) and time.
Max Costa introduced dirty paper coding (DPC) around 1983.2 His information coding is analogous to writing a message on a dirty piece of paper with lots of ink blots. Information written on the black spots can’t be deciphered and information is lost; however, if you were to write with white ink on the black spots, no information would be lost. Fig. 1 illustrates the concept.
Fig. 1: Representation of DPC
DPC coding estimates the channel characteristics and adapts the signal to compensate for the channel’s limitations. It is like a signal-processing form of pre-distortion.3 DPC uses spatial codes to effectively beamform, constructively interfering with specific users and destructively interfering with other users. The information is directed to a specific user at a specific location in space.
The DPC coding technique is different than phased-array beamforming, wherein the phase of each element in an array is adjusted to steer an antenna beam to one specific location. MU-MIMO is not steering the beam. The radiated beam width is not affected. Knowing the channel allows signal-processing directivity to pass the correct information to the correct recipient.
Massive MIMO
Massive MIMO reference schemes use a very large number of antennas, typically in the hundreds to even thousands. Under typical conditions, the capacity of a massive MIMO system increases with the number of antennas.4 The massive nature of the system yields other benefits. With so many channels pushing out information, it is not necessary to broadcast from any one antenna at very high power.
Conceivably, a massive MIMO system can radiate less than 1 W per channel and effectively support the same range with increased capacity compared to a traditional macro base station. The reduced power-out per antenna eliminates the need for expensive power amplifiers and complicated linearization techniques; standard Class AB amps with minimal to no linearization are sufficient.
Massive MIMO is best suited for operation in the centimeter- or millimeter-wave bands. At higher frequencies, the wavelength is short. The antenna size is small and the required spacing (half-wavelength at minimum) allows for massive arrays in a physically small space. The transceivers must also be very compact and highly integrated to facilitate the total number of channels.
Massive MIMO effectively circumvents a couple of traditional limitations with high-frequency broadcasts: propagation loss and power devices. High-power amplifiers are difficult to implement and are expensive at higher frequencies, which make their use in traditional base stations unattractive. In a massive MIMO application, the total radiated power is distributed across all elements in the network, so the radiated power at each element is low and does not require a high-power amplifier. With so many elements, the total power in the channel can be high enough to overcome the propagation loss while still having each element at a relatively low output power.
Although there is no theoretical limit to the capacity increase with additional elements, there is a practical one. Two real-world limitations affect system scaling: processing capability and channel integration. The processing power to solve the channel matrix H grows exponentially with antenna elements. You would need computationally stout processors to handle the calculations and small, efficient radio-frequency (RF) transceivers to support the additional elements.
Conclusion
The MIMO system is a viable solution for increasing capacity through increased data rates or for an increased number of users compared to a traditional single-channel link. The system thrives in a fading, multipath environment, taking advantage of the environmental coding imprinted on the information as it propagates through the channel. Implementing MU-MIMO and massive MIMO architectures support an arbitrarily large number of users. Expansion is limited by system cost and complexity to incorporate more RF transceivers and more powerful processor demands.
The MIMO technique is gaining traction in sub-6-GHz telecommunication bands. It is expected to be the primary architecture for higher-frequency millimeter-wave bands, in which a distributed antenna network is advantageous both from a radiated power-per-element perspective and achievable from an antenna-sized perspective.
References
1 Lina Jin (2011), “Scheduling, Spectrum Sensing and Cooperation in MU-MIMO Broadcast and Cognitive Radio Systems.” Retrieved from: https://www.escholar.manchester.ac.uk/jrul/item/?pid=uk-ac-man-scw:160486.
2 Max Costa, “Writing on dirty paper,” IEEE Transactions on Information Theory 29 (3) (May 1983), pp. 439–441, May 1983.
3 Andrea Goldsmith, Syed Ali Jafar, Nihar Jindal and Sriram Vishwanath, “Capacity Limits of MIMO Systems,” IEEE Journal on Selected Areas in Communications 21(5) (June 2003), pp. 684–702.
4 Matthias Stege, Peter Zillmann and Gerhard Fettweis, “MIMO Channel Estimation with Dimension Reduction,” 5th International Symposium on Wireless Personal Multimedia Communications (October 2002).
Related articles
The MIMO advantage, Part 2: expandable capacity in MIMO systems
Advertisement
Learn more about Electronic Products MagazineTexas Instruments
