The question now is, how can >10x performance improvements be achieved? While a number of approaches are being considered, the likely gains are marginal. However, there is plenty of ore left to be mined in the vein of radio system design by more fully using the dimension of space — in fact, at least 10x in the immediate future and a lot more in the long term. And, the mining of space need not wait for new innovation. It requires merely vigorous application of MAS technology that's already well proven.

Current wireless networks employ comparatively blunt instruments for the distribution of radio energy in physical space (Figure 2). As illustrated in the signal-pattern diagram of Figure 2, this approach creates vast amounts of waste in the system. Power is distributed where subscribers aren't and self-interference is created that degrades signal quality.

An approach using MAS software, in contrast, takes control of the space dimension and puts radio energy where it's required (Figure 2). MAS software drives an array of two or more antennas on the client device, the base station, or both, leveraging the principle of coherent combinations of radio waves to create a focus of transmit energy (or receive sensitivity) on the intended recipient (sender) and the absence of energy (sensitivity) on sources of co-channel interference. MAS-enabled devices can take advantage of a number of possible gains from using multiple antennas, including link budget improvements from diversity and combining gains, along with client data rate and overall network capacity benefits from active interference mitigation and spatial multiplexing.

Figure 3 illustrates how these different categories of MAS benefits add up to improve the overall performance envelope for wireless systems. Figure 4 shows how MAS software fits into common base station and client device architectures.

For base stations, the most common approach is to add a standalone, off-the-shelf DSP platform on the base station modem card to perform MAS processing. FPGAs can be used in theory, but are typically not in practice, because they reduce algorithm-upgrade flexibility once the equipment is deployed (an important capability to maintain). Program memory requirements for MAS software are negligible, but appropriate capacity for real-time signal processing must be reserved — on the order of 200 to 300 million CMACs per second for the suite of algorithms at the field's current state of the art. On client devices (such as handsets or PCMCIA form-factor data cards), the most common approach is to run the MAS algorithms on an embedded DSP core as part of the digital baseband ASIC. Here again, operation on a general-purpose signal-processing platform rather than hard-coding into dedicated gates is important to maintain the flexibility for upgrades over the life of the device.

Previous 1 2 3 4 Next

RSS    Save to Del.icio.us  Digg This