With the introduction of 3G wireless systems operating in the Universal Mobile Telecommunications System (UMTS) band, mobile operators saw the need to move to remote radio-head architectures to improve coverage and reduce operating costs. At 2100 MHz, the losses in RF power over traditional coax cable becomes prohibitive, so digital interfaces were required between the baseband processors at the bottom of the tower and the radio electronics at the top of the tower.

To foster competition in the radio-head market, mobile operators have demanded the standardization of these digital interfaces. The wireless infrastructure manufacturers responded by developing not one but two standards, the Common Public Radio Interface (CPRI) and the Open Base Station Architecture Initiative (OBSAI). Both CPRI and OBSAI specify how baseband data for individual carriers or channels are mapped into a common frame structure for a multitude of wireless technologies, including W-CDMA and WiMAX.

Current implementations for W-CDMA can transport this data over single-mode fiber-optic cables operating at bit rates of 3 Gbits/s. Thanks to the storage-area networking (SAN) market, low-cost fiber-optic transceivers are now available at rates of 4.25 Gbits/s and below. Many FPGA suppliers are also introducing low-cost families of products with serializer/deserializer (SERDES) interfaces able to operate at these rates.

But when 4G technology such as Long-Term Evolution (LTE) is introduced, this fiber-optic capacity will need to increase to 6 Gbits/s due to higher channel bandwidths (20 MHz versus 5 MHz for W-CDMA) and the use of multiple-input multiple-output (MIMO) technology (up to four antennas versus two-antenna diversity for 3G).

This change will present serious cost implications for the wireless industry, requiring FPGAs that are more expensive to handle the higher SERDES rates as well as fiber-optic transceivers, which can exceed the cost of the power amplifiers. This is not good news for mobile operators, who after all want to spend their capital expenditures on wireless, not wireline, technology.

Signal-compression technology is stepping in as a viable solution to this problem because it can enable 4G signals to be carried across the existing fiber-optic infrastructure deployed for 3G systems. Current signal-compression techniques can deliver truly lossless transmission, so there is no impact to signals sent across the fiber-optic link. In fact, in the uplink direction, the signal-to-noise ratio is quite low, allowing near lossless compression methods to be used.

With such compression techniques, the losses can be modeled the same as any other noise source as long as the noise level can be controlled and the spectrum of the compression noise is white. Additionally, to minimize the impact on latency, when transmitting over fixed-bandwidth fiber-optic links, there are compression algorithms that can adapt to maintain output bit-rate requirements to reduce the amount of buffering. And, the availability of a low-complexity compression algorithm enables them to fit into FPGAs in systems already in the field.

Signal-compression technology can also transform and simplify radio-head architectures and management. Take, for example, current designs where the power amplifier, digital pre-distortion processor, crest factor reduction processor, transmit digital-to-analog converter (DAC), and receive analog-to-digital converter (ADC) all operate on an operator’s entire block of spectrum.

The data is channelized into individual carriers in the digital domain to reduce the bit rate over the fiber-optic cable. By channelizing in the radio head, the baseband processor must inform the radio of the channel configuration, not just at initialization, but each time capacity is expanded or the system is upgraded. This requires complex and often proprietary management and control functionality to be built into the CPRI and OBSAI standards, limiting the stated goal of multi-vendor interoperability.

If signal compression is applied to broadband spectra to enable it to be sent directly over fiber-optic cables at economical bit rates, then radio-head architectures and management can also be greatly simplified. By co-locating the digital downconverter/upconverter (DDC/DUC) and channelization with the baseband subsystem, these functions would now be within the domain of a single vendor, eliminating the need for cross-vendor control and management traffic over the fiber link.

Unlike channelized architectures, broadband radio architectures that employ signal-compression technology are transparent to the type of signal, so the radio head also becomes future-proof and able to support further evolutions in the standards. In fact, signal-compression technology enables the goal of software-defined radios.

Allan Evans is vice president of marketing at Samplify Systems Inc. He holds an MSEE from UC San Diego and an MBA from Santa Clara University. He can be reached at aevans@samplify.com.