Programmable Wireless Systems Set to Change the Landscape
Written by Paul Dillien, Principal, High Tech Marketing
Published in EE Times
The widespread adoption of smartphones has driven a rapid change in mobile communications. The various wireless technologies known collectively as 3G were optimized to be efficient at handling voice traffic. However, smartphones have shifted this dynamic so that data traffic in various forms now dominates. Add to this that one-in-three tablet computers feature cellular connectivity, and it becomes clear why data revenues for US cellular operators now exceed voice revenues.
While revenues may be roughly comparable, the number of bits that must be transmitted is significantly different. Data traffic is estimated to consume around 85% of the total traffic and is growing at 20% per year, so operators must optimize their networks to accommodate this shift. Fortunately, this transition was anticipated: The technology known as Long Term Evolution (LTE) has been specified by the 3rd Generation Partnership Project (3GPP), and new equipment is rolling out around the world.
LTE has been designed to be far more efficient at transporting bits; it approaches the maximum information rate that can be achieved within a given bandwidth, as defined by Shannon's Law. The original LTE has been superseded already by LTE-Advanced, and "5G" is slated for the end of the decade.
To achieve an even higher throughput, the operators have a number of options. They are allocating more bandwidth to LTE to provide fatter pipes. The target maximum bit rate for a 20 MHz bandwidth is 100 Mbit/s for download and 75 Mbit/s for upload. Operators are also "re-farming" spectrum from older technologies by changing out the equipment to support LTE technology. Globally, there are more than 40 different channels specified for LTE use, which is a major challenge for equipment vendors wishing to serve all the markets. However, there is a finite amount of spectrum, and operators must live or die by their available resources.
Even with the enhanced spectral efficiency from LTE, it became clear that more needed to be done. It was realized that simple networks are no longer appropriate. The older technologies rely on a small number of higher-power cellular towers to provide coverage over a wide area. These are connected to macro base stations, which support multiple users simultaneously.
There are several issues with this setup, including the fact that the signal level reduces by the inverse square of the distance. As a result, the signal-to-noise ratio (SNR) worsens as you move away from the tower. The result is that the data rate experienced by the user will be dependent on the distance to the nearest tower. Operators cannot boost the transmitted power, as this would create interference with adjacent cells. In any event, that would only affect the download data rate, but the users would still be limited by the much lower power that their equipment can output. A further consideration is that -- since more of the available bandwidth must be dedicated to each user -- the macro base station can support fewer simultaneous users, which is obviously detrimental to the operator.
Exacerbating the problem, a lot of the newly released spectrum is at much higher frequencies than the original 700 to 850 MHz. The higher frequency of up to 2.7 GHz has two effects. First, the signals are more easily absorbed by buildings and trees, so that the effective range of a tower is reduced. Second, this absorption means that signals do not penetrate intobuildings, which is where much of the traffic originates.
One solution to these problems is for the operator to reconfigure his cell boundaries and to install more macro base stations that are designed to service a small-cell area. This is not a practical solution. The costs associated with site acquisition, equipment installation, and obtaining planning consent for tall masts are significant. Planning restrictions present a significant barrier in many localities. The unsightly towers are a detrimental change to the landscape, and adverse public reaction to them should not be underestimated.
Again, this scenario has been anticipated by 3GPP, and the LTE specification includes the provision for operators to move to a new cellular arrangement known as heterogeneous wireless networks (HWN). A typical system will still employ a macro base station for wide area coverage, but it also uses a range of small cells to supplement the service.
Vendors are producing micro and metro cells, along with picocells and femtocells. Each has different characteristics in terms of effective range and the number of simultaneous calls that are supported. The small cells sit inside the coverage area of the macro system, and act as off-loads for the macro. Excellent examples of sites that have heavy usage in a very local area are shopping malls and airport terminals. These are ideal for small-cell offloads. In the central business districts of cities, there is heavy demand during the working day contrasted by light traffic at night. The dynamic nature of the load allows the small cells to be powered down to save energy when not required.
Small cells range in physical size down to a WiFi router box for femtocells used in domestic or small-office/home-office (SOHO) applications. Clearly, in this case there is no requirement for planning consent, but picocells and above can be located in public or outdoor locations. Innovative solutions to reduce the visual impact of these units include attaching cells to street lighting poles or inconspicuous places on the walls of buildings.
To overcome the burdensome, lengthy, and expensive network planning exercise that historically was needed for cellular networks, HWNs can use a system called self-organizing networks (SONs). Here, the cells have intelligence that detects the frequencies being used by the macro and adjacent small cells and selects non-interfering channels. This self-configuration can involve communication and negotiation with other cells to optimize the setup. If a cell fails, SONs can also adapt to provide a self-healing system. The combination of HWN and SON provides increased reliability, improved spectrum efficiency, and increased coverage.
The small cells, however, need to be designed with both intelligence and an extremely flexible RF system. The option of designing and building an ASIC to perform the required logic no longer makes sense. This is because the NRE costs are prohibitive, and also because the 3GPP specification is regularly updated with new releases, making any fixed functionality obsolete.
As a result, FPGAs are the preferred solution. Moreover, with the incorporation of high-performance processors into FPGAs, these solutions become even more attractive. The logic functions in the FPGA take the data to be transmitted and manipulate it into a format suitable for transmission or reception. This might be pre-filtering the signal to shape or limit the bandwidth, turbo coding to provide resilience in the presence of noise, and CRF (crest factor reduction) to reduce the min-to-max signal level.
Another popular function is to include digital pre-distortion (DPD). DPD can reduce the overall cost by using a cheaper, but lower-performance, RF amplifier. The 3GPP specification calls for a tight specification on the amplifier, but DPD is a method of counteracting the imperfections to provide a closed-loop control that linearizes the system. This requires complex interactions between the RF and logic as the linearization is dependent on factors such as frequency and RF signal level.
Selecting an FPGA that features both on-chip dual ARM processors and high-speed serial transceivers allows an even greater level of integration along with cost and power savings. For example, the Altera Cyclone V SX range includes transceivers running up to 3.125 Gbit/s that can interface via fiber or co-ax to support the backhaul function. The ST versions of Cyclone V push this capability up to 5 Gbit/s. The ARM cores can control the data flow and -- working with a suitably sophisticated RF transceiver -- can be used as part of a self-organizing network.
A recently announced RF transceiver chip from Lime Microsystems gives the triple benefit of being highly integrated, low power, and very cost effective. The LMS7002M is fully programmable and is a second-generation field-programmable radio frequency (FPRF) device. There are dual transceivers, which cover the entire 3G and LTE spectrum, making them ideal for small cells. In the case of the transmit function, the digital bit stream is loaded as in-phase and quadrature data (I&Q) and first processed by powerful on-chip DSP functions to minimize the analog/RF distortion. The transmit DACs produce an analog signal that is further filtered to sit within the programmable RF modulation bandwidth, which can be selected from a range 0.1 to 108 MHz.
The next process is a programmable gain stage before the signal is mixed with the local PLL frequency to produce a modulated RF frequency. The PLL is again fully controlled by configuration data loaded into the device to support any RF in the range 50 MHz to 3.8 GHz. The RF signal passes through a further programmable gain stage before being passed to an external amplifier. The chip covers all the LTE frequency bands and solves the problem of designing a universal radio.
The receive function is equally programmable, with the option to use any of three low-noise amplifiers to accept the RF signal, which is mixed, filtered, and finally output as an I&Q bit stream.
The LMS7002M also features a received signal strength indicator (RSSI), which provides an instantaneous indication of the power at the selected frequency. This is ideal for SON, because it can be combined with a sweep through the frequency band of interest to identify which channels are being used and where there are opportunities to transmit without causing interference to existing users.
Combining an FPRF with an FPGA that has on-chip processors and high-speed serial transceivers provides a powerful solution to small-cell design challenges. The processors can interface with the backhaul using an FPGA core that implements a Common Public Radio Interface (CPRI) or Open Base Station Architecture Initiative (OBSAI) standard. The processors can also work with the logic in the programmable fabric to configure and format the data into the appropriate duplex schemes. This might be either frequency-division (FD) LTE or time-division (TD) LTE, which are two different standards of 4G LTE technology. The logic can also include hardware-accelerated filtering and shaping, as previously discussed, as well as being configured to create a JESD207 interface with the LMS7002M for data transfer between the chips.
The ARM cores can also be used to fully program the performance of the FPRF using an SPI interface. The program for each element of the RF device is performed by an address and data transfer that is both simple and fast. The processor can, for example, tune the receivers across the chosen spectrum with a precision below +/-25Hz, or can scan rapidly through by sampling at, say, 1 MHz intervals. The FPRF uses a reference clock in the range 10 to 52 MHz that is used to drive the on-chip PLLs to the required RF frequency.
The RSSI data at each point is used to map the spectrum usage. Once the SON has settled on a channel selection that will not cause interference, the ARM cores can program the Tx and Rx frequencies, the bandwidth, the gain, and the filter characteristics for operation.
The LTE specification features an additional technique to increase the data transfer rate. This is called multiple-input/multiple-output (MIMO). This is a complex scheme that uses two or more antennas that are physically separated. A system using two antennas is termed a 2x2 MIMO, which is directly supported by a single LMS7002M. The LTE also includes provision for 4x4 and 8x8 MIMO, which can be realized using two or four FPRF devices, respectively.
MIMO techniques improve the spectral efficiency and achieve a diversity gain that improves the link reliability. It is expected that MIMO will become an important addition in the operators' struggle to meet the growing demand for data service.
The combination of rapidly increasing data demand on networks and the continual evolution of the standards is a significant challenge for operators. They need to exploit every technical advantage to boost the performance of their networks in order to ensure that their customers remain satisfied.
Can you remember the days before cellphones? What do you think our usage model will look like in, say, five or ten years' time?