Mobile WiMAX, with the release of 2×2 MIMO chips in 2008, gives WiMAX a lead of two or so years on its major competitor—the 3GPP’s LTE. However, 3G cellphones using 3GPP UMTS technologies, extended to higher speeds with HSPA, is widely used in handsets in many countries. In North America, 3GPP2 CDMA2000 and EV-DO are widely used, but these are likely to be replaced over time by LTE and to some extent WiMAX.

LTE provides a smooth extension of the elaborate service provision system established with GSM. This includes the use of SIMs as the method of enabling a given handset to provide a service for a particular customer. This is a highly successful system for companies whose primary focus is cellphone handsets.

SIMs and customer identification systems

A wireless telephone service can only operate when one device is uniquely identified as belonging to a given number, and so is linked to that customer’s phone number. The SIM is a key aspect of this service, since any SIM works in any phone, and the mobile carrier needs only program a SIM, and maintain communications to it whenever it is in use, without needing to be concerned about exactly which physical handset it is plugged into. In order for the SIM to allow the phone to operate at all, an encrypted exchange takes place between the SIM and the carrier’s central Authentication Centre. This changes data stored in the SIM, which means that in principle a SIM could be copied, but that only one of the two or more copies would continue to operate in the network.

LTE, like GSM and UMTS, will require the use of a SIM. This is convenient for handsets, but makes the service difficult or impossible to operate with any other device, such as a laptop computer, which has no SIM interface.

WiMAX requires no SIM or other such hardware token—so whatever authentication methods are used to identify a customer’s device, they will be configuration items which can be easily entered into different types of device. As such, the one WiMAX device may be configured to use one of a set of customer identification settings, enabling it to be easily used for multiple different WiMAX networks in different locations, or within the same network for different customer identities.

Interworking with 2.5G and 3G networks

When used primarily for a handheld device with cellphone functions, LTE is at a considerable advantage in that when the device is beyond range of LTE base stations, it can fall back to 2.5G and 3G services, assuming it has the requisite radio technologies, with potentially seamless handover.

In principle, it would be possible for a WiMAX device to do so as well, but its model of service provision is not as close to 2.5G or 3G as LTE’s. For instance, any WiMAX device using 2.5G or 3G will require a SIM.

Mixed voice and data

Both WiMAX and LTE are based on IP packet-based data carriage, in contrast to 3G and earlier cellphone systems which were designed to accommodate circuit-switched communication. The LTE specification comes from a background of cellular telephony. It would not be surprising if LTE’s QoS arrangement and more elaborate upstream and downstream arrangements make it more suitable for voice. LTE, being developed while mobile WiMAX is being deployed, benefits from another two years or so experience and innovation in this rapidly changing field.

Uplink power efficiency—OFDM versus SC-FDMA

Both WiMAX and LTE use OFDMA for the downlink and so have broadly similar performance, for any given RF bandwidth and set of conditions. By contrast, the modulation techniques for their uplinks are entirely different.

WiMAX (including Mobile WiMAX) also use OFDM for uplink, while LTE uses a new technique—SC-FDMA (Single Carrier Frequency Division Multiple Access). Hyung G. Myung’s site [hgmyung.googlepages.com/scfdma] provides a detailed description of SC-FDMA.

SC-FDMA resembles OFDM in many respects, but involves using a Fourier transform to convert OFDM’s separate sub-carriers on separate frequencies into a different, time-domain, form. With OFDMA, several handsets can transmit upstream at the same time, within the same set of 512 sub-carriers, with each handset transmitting data on different sub-carriers. This reduces the frequency diversity of each handset’s signal, but enables several of them to transmit upstream at the same time. SC-FDMA also enables simultaneous upstream transmissions, and the receiver is able to separate the separate components from each handset after suitable mathematical transformation.

SC-FDMA, as it is planned to be implemented in LTE, results in a transmission properties with some of the characteristics of both OFDM and DS-CDMA (Direct Sequence Code Division Multiple Access). These are two completely different modulation techniques, and SC-FDMA can be considered as a novel, but useful hybrid of the two.

Both SC-FDMA and OFDM can be used with fast-moving mobile devices, with some compromises in the total achievable data rate. Part of the problem in accommodating fast-moving devices is the Doppler shift resulting from the movement, including from the mobile device’s point of view. Another is coping with the rapid changes in propagation environment, such as those due to multipath reflections.

SC-FDMA can also be used with MIMO, as it will be in LTE. As we discuss below, SC-FDMA will also be the uplink modulation scheme of the future WiMAX 802.16m system.

Peak to Average Power Ratio (PAPR)

The primary attraction of SC-FDMA over OFDMA is the former’s significantly reduced Peak to Average Power Ratio (PAPR). ODMA’s output waveform is the sum of hundreds of sub-carriers, each of which is composed of uncorrelated levels of sine and cosine signal. The result of summing a large number of small uncorrelated (essentially random, with respect to each other) numbers is that most of the time, the sum is relatively small too. However, occasionally, when many of the sub-carriers are strongly positive, or strongly negative, the sum is a much larger value than usual.

If these values were numbers in a digital computer system, this would present no problems. However the digital OFDMA signal is converted to an analogue signal by a Digital to Analogue Converter (DAC), and amplified to drive the transmit antenna. The signal propagates through space, being very strongly attenuated and mixed with interference and other noise, where it arrives at the receive antenna. There, it is amplified again and then fed to a high resolution, high speed, Analogue to Digital Converter (ADC). Now in a digital form, the receive signal is analysed by a Fourier Transform operation and many other processes to recover the transmitted data.

Because OFDM signals have a high PAPR, the linear amplifiers which must be used to drive the transmit antenna, and in the receiver, must have a large headroom compared to the average value of the signal they are handling. It is possible to set the amplifier’s power supply so that the very highest peaks, which are quite infrequent, will be clipped—but the more clipping which occurs, the greater the distortion, which adds noise to the received signal and makes it harder to discern the finer increments of modulation: four or more levels of sine and cosine in each sub-carrier.

Signals with a high PAPR are most problematic in battery operated transmitters. The RF amplifier must have a relatively high positive and negative power supply in order that its transistors can lift and drop the output waveform linearly to follow the extreme, but relatively infrequent, peaks of the OFDM signal. While the average current used by the amplifier may be quite low, and its average output level also quite low, all this current is drawn from these high voltage supplies.

For a signal with a low PAPR, the same average signal level can be achieved with an amplifier with much lower power supply voltages, because the signal does not have high peaks.

Consequently, to achieve a given average power output, an amplifier with +/- 12 volt supplies might be required for an OFDM signal (allowing for some low, but acceptable, level of noise due to the very highest peaks being clipped) or with the same amplifier with +/- 4 volt supplies for an SC-FDMA signal with a much lower PAPR.

Since the average current drawn by both amplifiers is the same, and the power drawn is the current multiplied by the power supply voltages of the amplifier, in this example, the SC-FDMA transmitter only uses a third of the power of the OFDMA transmitter.

Inefficiency in transmitter amplifiers presents few problems in base-stations, but is a major determinant of battery life in a cellphone or other mobile device. Consequently, considering all the tradeoffs, for LTE it was decided that SC-FDMA was a better choice for the uplink modulation technique.

This is likely to result in LTE being more suitable for handheld devices then WiMAX. This advantage of lower power consumption would not be so important in a device with a bigger battery, such as a laptop—and of course it is no advantage for a mains-powered fixed WiMAX service.

WiMAX chipset power consumption test

In late 2008, two mobile WiMAX chipsets were carefully tested for power consumption for both voice and data patterns of usage, connected into a test jig which simulated a wireless connection to a base station [See http://www.wimaxtrends.com/2008/10/low-power-wimax-chipsets-accel.html]. All major chipset vendors were invited, but only two took up the offer.

The GDM7205 chipset of GCT Semiconductor compared somewhat favourably with the consumption of chipsets in current mass-production EV-DO and HSDPA 3G handsets, including a Palm Treo, an Apple 3G iPhone and a RIM Blackberry. The ALT2150 chipset from Altair Semiconductor performed markedly better.