Site planning for Mobile WiMAX

Traditional cellular base station approaches to tomorrow’s networks are evolving

BY J.R. SANFORD
Laird Technologies
Carlsbad, CA
http://www.lairdtech.com

Traditional cellular phone radios can work over a large range of input signal. Close to a base station, signal levels are in the region of –40 dBm. Although this would not be the design aim of the system, the phone still will have acceptable quality at –113 dBm on the edge of a cell.

The only advantage to significantly higher power is to combat interference. In the early days of cellular planning, the focus was on coverage rather than interference or network capacity.

Today, however, most cellular systems are interference limited and not range limited. As a result, many cell-phone systems required an expensive fork-lift upgrade to accommodate the rising interference constraint.

Often, omnidirectional antennas were replaced by directional antennas. In some cases, directional antennas were replaced by other directional antennas with better front-to-back ratios. Over time, the beamwidth and pattern shapes were chosen to provide a good blend of range, capacity, and interference rejection.

Signal-to-noise ratio

This inefficient optimization process may be repeated for WiMAX deployments. The difference between the WiMAX system and traditional cellular systems is usually not considered in the planning phases.

New systems are planned in the same way traditional cellular phone systems were planned. However, WiMAX works much differently.

If the coverage is adequate and signal-to-noise ratio (SNR) is above about 9 dB in a traditional cell-phone system, users can make a crystal-clear call and be satisfied. If the customer has a 20-dB SNR, the call works just the same.

However, in a WiMAX-based system, a 20-dB SNR will allow the customer to receive a signal in a modulation format that provides a very high data throughput. If a 9-dB SNR is received, the link is much slower because WiMAX systems use an adaptive modulation that is based on SNR. The analogy for human communications is simple.

If it is quiet and the speaker can easily be heard, then the speaker can speak quickly and be understood. However, if it is noisy and the speaker is far away, the user needs to yell and speak more slowly. The communications link is inefficient.

Dealing with echoes

Another fundamental difference between WiMAX and traditional cell-phone networks is over-the-air modulation. Traditional cellular systems such as GSM use a method that can be considered a signal carrier system.

The total RF spectrum is divided up into frequency channels and each user uses a channel for communication. Multipath echoes limit the speed at which one can communicate, similar to the echoes caused by people yelling across a canyon.

In order to communicate clearly one needs to speak slowly so as to not be confused by the echoes. WiMAX uses OFDM modulation that eliminates the echo problem.

The conversation is broken up into a number of slow conversations, each using a different tone. Since each individual conversation is slow, the echoes are not a problem.

Each conversation is at a different tone so they don’t interfere with one another. While this might not be practical for individuals communicating across a canyon, it is easy to implement in circuitry.

While the goal for traditional cellular telephone systems was to provide a 9-dBi SNR or better everywhere with no dead spots and low multipath, the WiMAX system would like to have the best average SNR over the coverage area with no dead spots and much less concern with multipath. With these fundamental differences between WiMAX and traditional cellular it seems unlikely that the same planning methodology and antenna hardware would be appropriate.

Determining the best configuration can be daunting. Every deployment is different and the number of variables is huge. The variables include the number of sectors used, the frequency reuse scheme, the RF propagation environment, tower heights, and background interference level. An advanced optimization process together with propagation models help us explore some of these issues.

Simulation foradvanced optimization

The following simulation uses the Cost 231 Model—also called the Hata Model PCS Extension. This radio-propagation model extends the Hata Model and Okumura Model to cover a wider range of frequencies, such as those common to WiMAX.

The full 3-D radiation pattern of base-station antennas is read and the carrier-to-interference ratio (C/I) throughout the cell is viewed. The C/I ratio is equivalent to the SNR, where the signal of interest is the OFDM carrier and the interference is coming from some other base station that uses the same carrier.

The height of the tower is considered in the model. The optimized variables are the pointing direction of the antennas, the number of antennas on the tower, and the height of the tower. The goal is to find a site plan that provides the best possible total system capacity, a task that is highly coupled to the average C/I in the cells.

Network optimization

First, look at the effect of base-station antenna pattern shapes. A frequency-scaled premium UMTS base-station antenna is compared with a 72° 3-dB beamwidth with a WiMAX antenna that has a 90° 3-dB beamwidth.

While the UMTS antenna has slightly more gain, the WiMAX antenna has much lower backward radiation. The UMTS-style antenna has a pattern mask that meets the ETSI SS1 pattern specification while the WiMAX antenna has a pattern that meets the ETSI SS3 specification.

Channel reuse is not recommended within the seven-tower lattice considered in the optimization and the manufacturer optimizes of best total data throughput. The result of the optimization is shown in Fig. 1 .

Fig. 1. The result of manufacturer optimization of total data throughput. Left: Best-in-class UMTS-style CS2 antenna with 18-Mbit/s downlink. Right: Laird’s 90 CS3 802.16d antenna with 27-Mbit/s downlink. The signal levels are about the same, but the interference is reduced.

Each tower is denoted by a letter. Each identical letter is an identical tower configuration. The color indicates the C/I ratio and the capacity at each location is computed from the plot in Fig. 1. The average throughput in a 16.5-MHz half-duplex channel is shown below.

Interestingly the optimizer does not choose the standard three-sector configuration so common to cellular base station but rather a four-sector configuration for both antennas. We see that the more stringent pattern mask results in a significantly better average C/I and hence a better base station capacity. More insight can be gained by looking at the coverage diagram.

Fig. 2. The lower backlobes of the WiMAX antenna provide for cleaner transition zones between sectors. Left: UMTS-style CS2 antenna with 18-Mbit/s downlink. Right: Laird’s 90 CS3 802.16d antenna with 27-Mbit/s downlink.

Figure 2 shows the lower backlobes of the WiMAX antenna provide for cleaner transition zones between sectors. Mobiles in schemes using the lower-performing UMTS style antenna can get “confused” about what sector they are due to the formation of microcells caused by the antenna backlobes or an improperly specified beamwidth. This can make handoff difficult for the device.

This analysis generates the question, what are the correct antenna beamwidth and sidelobe specifications for a WiMAX antenna? As is often the case, the answer requires a set of engineering tradeoffs.

Consider the case where the operator is spectrum limited and wants to get the most out of seven available channels to be used among the seven-tower configuration that is periodically repeated over the landscape. By choosing only seven channels the operator has upped his maximum burst capacity. However, it is not obvious that a reuse pattern can be found that will provide a sufficient C/I to operate.

This is clearly a challenging problem. We allow the optimizer to select the optimal antenna as well as the configuration. The antennas that can be used are the WiMAX 60° antenna, the WiMAX 90° antenna and the scaled UMTS 72° antenna.

The optimizer determines that the best configuration when using the 60° antenna is a four-sector scheme. Each beam pattern is used four times within the seven tower periodic layout. As such, we have a four-time spectral reuse scheme.

Unlike the previous designs that allowed primarily 64 and 16 QAM to be used over the area, this scheme blends the full range of modulations that WiMAX supports. The average C/I can only support 14 Mbits/s on average in the half-duplex 16.5-MHz channel. However, with the four-time reuse factor this increases to an impressive 58 Mbits/s. The liability of such a deployment would likely be the reliability. The scheme does not offer much fade margin.

In the future

Going forward, the antenna design will be coupled directly into the network optimization process. This will allow companies to design antennas simultaneously with the network optimization, ensuring the operators can realize the full network capacity that is available.

For simplicity, thus far cases that assume single antenna input on the transmit end with single antenna output on the receive end (SISO) have been the focus. The analysis approach shown here is also applicable to multiple imputer multiple output (MIMO).

The change is in the channel model. The Cost 231 propagation model would require some adjustment since a MIMO system would have better propagation characteristics with lower fading variance. The general conclusion remains largely the same. ■