Smart antennas embed in wireless nets

n today’s code-division multiple access (CDMA) networks, boosting network capacity is traditionally done in one of two ways: by adding additional carrier frequencies or increasing the cell site density in the network. Adding CDMA carrier frequencies has associated costs and deployment difficulties. Issues with the reliability of carrier-to-carrier handoffs force operators to deploy additional carrier frequencies in clusters, or in some cases network-wide. Such deployments are expensive and time-consuming, especially given that network capacity problems typically occur in localized hotspots of the network. Thus, deploying additional carrier frequencies is often an inefficient solution. Further, it’s impossible to add carrier frequencies when available spectrum has been exhausted.

An alternative option is to add cells to networks to offload traffic, a technique known as cell splitting. However, the logistics of acquiring real estate, obtaining zoning approval and covering the additional maintenance costs of new sites often renders this solution unattractive, if not impossible. Further, in dense urban areas cell radii may already be so small that cell splitting is not viable. Cell splitting works when the cells are far enough apart that a mobile that is served by one cell does not seriously impact neighboring cells. When cells become too small a mobile will create too much interference in neighboring cells.

Smart antennas offer yet another method for boosting capacity in CDMA networks. Until recently, the use of smart antennas had been limited to military applications. But performance gains in cost-effective signal processing and improved algorithms have helped smart antennas find a new home in the commercial wireless industry.

Wireless systems typically are interference limited. Each signal that is broadcast from or received at the mobile set interferes with signals for every other mobile set in the vicinity. Likewise, transmission signals from one basestation interfere with those from neighboring basestations. In essence, each new call introduces interference to the system. This phenomenon continues until the system has reached the maximum allowable noise level. Once the noise level is maxed out, forget about adding incremental calls.

However, more users could be added if each one contributed less noise. Smart antenna technology reduces the amount of noise per user, thus improving system capacity. Embedded smart antennas function by calculating and forming focused beams for individual mobile sets. This technology provides capacity improvement in wireless systems through the more economical use of basestation signal power. The synthesized beams are calculated and adjusted over time based on an estimate of each mobile set’s location, and by taking into account environmental conditions such as multipath. By using specific individual beams designed to be more efficient than the standard sector-wide beam, smart beam synthesis technology generates a significant reduction in interference and thus an increase in system capacity. Some industry estimates suggest that the use of embedded smart antennas will enable call-capacity increases of between 200 percent and 300 percent in CDMA networks.

It’s helpful to understand the architecture of the baseband processing for embedded smart antennas that has been implemented for CDMA 2G and 3G. The general architecture consists of one multielement antenna array per sector, which provides both receive (Rx) and transmit (Tx) paths. The basestation’s radios amplify, down-convert and demodulate the RF/IF signal from each array to obtain a sampled version of the baseband signal. These baseband signals are then moved to the smart antenna card (SAC) and the basestation’s uplink processing circuitry.

The SAC’s software and hardware use the baseband signals they receive from each element of the antenna arrays to determine the optimum uplink and downlink configurations for the basestation. The SAC uses this information to determine the direction of the mobile set with respect to the basestation. From these parameters, the uplink and downlink beam-forming coefficients are established by software and hardware on the SAC. The SAC generates appropriate data and passes it on to the beam-formers, normally placed on the basestation’s channel processors.

The downlink/uplink processing structure of the basestation is modified to include beam-formers in the processing path. The carrier’s phase and amplitude must be controlled on each element of the antenna arrays in order to generate downlink beams. This is done by multiplying the in-phase and quadrature-phase baseband signals by the appropriate complex coefficient to yield the desired phase and amplitude relationship. These multipliers are the beam-formers.

Accurate beam-forming requires precise knowledge of the amplitude and phase relationships among all of the antenna elements and their associated hardware (cables, filters and the like). The SAC calibrates these paths by injecting and receiving signals into the RF portion of the system.

For each antenna port, the phase and amplitude relationships to an incident planar wavefront are obtained for the angles and frequencies of interest. This is done before deployment and stored in a table called the Array Manifold table. The Array Manifold table is essential for estimating angle-of-arrival information and computing the weight sets required for beam-forming in the desired directions.

For each antenna path, calibration weights must be applied to compensate for hardware variations such as cable lengths and environmental variations such as temperature. The phase characteristics of the antennas are altered by all of these effects and distort the resultant beam. To minimize this distortion, compensation is applied by means of an Antenna Characterization Table, which used in conjunction with the aforementioned Array Manifold Table, comprise the calibration process. A calibration signal is sent by the SAC to analyze the antenna paths to enable automatic calibration and ensure that the amplitude and phase characteristics of each antenna path are understood.

The SAC consists of both the hardware and software necessary to perform the mobile location and beam-formation process.

ASICs and field-programmable gate arrays (FPGAs) perform the reverse link channel despreading and channelization. These functions are needed to identify individual traffic channels so that they can be separately processed to determine their location and other characteristics for creating accurate beams. All traffic channels are processed so these components need to function on multiple traffic channels simultaneously.

A microprocessor provides the SAC’s data-processing capabilities for beam coefficient estimation, as well as inter-modular communication and embedded control tasks. The microprocessor also handles the low-level real-time tasks and hardware control operations.

A high-speed link (contained on a backplane within the basestation) provides the interface between the SAC and the channel processors. This interface passes along the beam-forming coefficients. The beam formers are used to generate the forward and reverse link beams for each channel.

Additional functionality that is not necessary for the implementation of smart antennas is added to the SAC to aid in testing and debugging. One of these is the Ethernet interface that provides communication between the SAC and an external PC. The PC will support a real-time data analysis program that is used to collect, display and store SAC data. This link also provides diagnostics and configuration support.

Baseband processing has been implemented in semicustom logic such as ASICs and FPGAs, as opposed to using digital signal processors, because it was determined that the performance of the fastest DSP (at time of development) could not approximate that of an ASIC or even an FPGA for this application. This is due to the parallel structures of semi-custom ICs as compared with the serial/iterative structures of DSPs.

These components have to work on a large number of traffic channels simultaneously. They require a huge number of complex multiplications and additions to perform the integration and the correlation between each traffic channel and the antenna array manifold. The sheer number of steps that have to be performed meant that components had to enable highly parallel structures. ASICs and FPGAs held a large advantage over components that functioned in a more serial fashion, like DSPs. For example, a vector multiplication (dot product) of two 16-element complex vectors can be performed in 13 ns using the ASIC technology. The DSPs we looked at could not approach that speed.

Raw magnitude and spatial data (the angle of arrival and time of arrival) are sent from the Reverse Link Processor to the microprocessor for analysis. Using additional information from the basestation’s channel card, an algorithm is developed in software to determine the best Rx and Tx beam-forming coefficients. Once these coefficients are determined, they are passed on to the beam-formers on board the channel processor cards. There is a beam-former (multiplier) per channel. This provides the beam-shaping capability. It’s here that the coefficients form the corresponding channel’s beam(s), through a complex multiplication or dot product operation.