A question comes often in wireless industry circles on what is more advantageous to use in wireless products: programmable devices such as FPGAs and DSPs or ASICs? This question, when applied to wireless infrastructure products, finds many wireless network operators siding with SDR (software defined radio), which uses programmable devices. They cite the ability to migrate with evolving wireless standards, thereby reducing capital expenditure associated with wireless base stations.
Equipment manufacturers, on the other hand, have been under severe pressure by the network operators to keep the cost of equipment low, which does not favor fully programmable products. This dichotomy has contributed to some tension between the two camps due to the large amounts of capital spent on wireless infrastructure products — in particular, base stations. The role of cost control in driving consolidation of both operators and vendors is well known.
To address this question, we need to take a close look at the underlying economic and technical trends that are affecting the wireless industry. One trend is the proliferation of smartphones, which is driving data traffic volume. USB dongles can quickly enable laptops and netbooks to connect to the wireless network — and soon, embedded chipsets would become standard edition in many such devices. This will provide impetus for mobile traffic data volume to double every year until 2014.
Some wireless operators have reported very dramatic increases: AT&T wireless reported a 5,000 percent increase in the last three years; O2 in the U.K. said data traffic doubled every three months in 2009; and Telecom Italia announced data traffic increased by 216 percent between mid-2008 and mid-2009.
To support the expected data traffic growth, new standards such as LTE are being developed. LTE and other 4G broadband wireless standards rely on large channel bandwidth (5, 10 or 20 MHz) and multiple antenna systems (MIMO), among other features, to provide high data rates. Often, these standards will be deployed in higher frequency bands, such as 2.5 GHz, where propagation losses are more severe. To enable higher capacity and compensate for the higher propagation loss, network operators are exploring a deployment paradigm that features hierarchical networks that include macrocells overlay with a microcell (or picocell) underlay.
This paradigm provides high data rates through small microcells deployed below clutter to limit interference and maximize capacity while supporting high-speed mobility applications on macrocells to reduce handovers, minimize dropped calls and increase service reliability. The underlay microcell deployment takes cell-splitting to its ultimate end and is expected to improve coverage as well as capacity. For the deployment model to be successful, it requires low cost and small size base stations in addition to low operational cost related to leasing space on poles and backhaul expenses. Low cost base stations can be achieved by using standardized, high volume components with low power consumptions, which points against fully programmable devices.
Yet wide-scale deployments cannot happen until the industry has made a conscious decision to adopt a certain standard. Infrastructure makes for a large portion of an operator’s capital budget; the technology itself has wide effects on the profitability of an operator as it is related to consumer devices and a myriad of other factors that directly impacts profitability in both the revenue and expense side of the accounting ledgers. Standards need to be stabilized and bugs worked out. Additionally, there is a need to address the continuous evolution of standards. For example, UMTS went through several releases to upgrade its theoretical capacity from 2 Mbps to about 40 Mbps, which included technologies such as HSDPA, HSUPA and HSPA+. Similarly, LTE Advanced is currently in the works and it features a full OFDMA uplink physical layer as opposed to SC-FDMA for LTE. Such evolutions take place over a short period of time — sometimes as low as two years — which is short in comparison to the time it takes to develop, test and validate equipment. The need to evolve existing infrastructure points against full ASIC implementation due to the limitation of adaptability.
To keep the cost of base stations low to enable a deployment model that can provide the capacity for future data traffic requirements while allowing evolution with wireless standards creates a dichotomy that can be resolved through a new generation of System on Chip (SoC) devices that combine aspects of programmable devices and ASICs. SoCs dedicated to communication systems combine “hardened cores” of fundamental building blocks common to a generation of wireless standards with programmable logic units. For example, the physical layer of 4G wireless standards that include LTE and WiMAX is based on OFDMA, which uses inverse fast Fourier transform (IFFT) function. IFFT/FFT functions can be provided in the SoC to accelerate the performance of the physical layer and reduce power consumption. Similar blocks can be provided for other key functions, such as channel coding (e.g. CTC), encryption (e.g. AES-128) and interfaces (e.g. CIPRI, OBASI), in addition to memory to reduce latency in signal processing by reducing the need to access off-chip memory blocks. Meanwhile, other functions, typically pertaining to higher layers of the protocol stack, can be implemented on programmable logic within the SoC which enables adaptability and evolution with the standard.
The use of SoCs in wireless base stations has been taking hold at the level of femtocells, which, as home user devices, necessitate a very low cost to enable a successful business model. At the other end of the spectrum, most of today’s macro base stations have relied on programmable devices such as FPGAs and DSPs to provide core baseband functionality.
As the trend toward microcell deployment increases, it is expected to see a rise in the adoption of dedicated SoCs to enable the business case for both operators and equipment vendors. It then becomes critical to targeting the right mix of hardened cores and programmable capabilities with sufficient integration to maintain low cost and low power consumption.
In conclusion, data traffic imposes new requirements on wireless network deployment paradigms that will necessitate adoption of new technologies to enable a profitable business case. This points to an increase in the use of SoCs that can provide the right mix of adaptability and cost advantage.
Frank Rayal is CTO at Telesystem Innovations.