It is now generally accepted that the exponential trend in wireless data traffic growth cannot be supported by existing 3G/4G network topology on either an economic or technical basis.  Further, there is general agreement that the existing “macro base station only” infrastructure cannot support the increasing demand for bandwidth since throughput degrades significantly over distance. The extreme popularity of smartphones posed enough of a problem, but now network operators must confront the arrival of tablets, ultra books and increasingly mobile work forces from enterprises large and small, as illustrated by the BYOD (bring your own device) trend. 

Notwithstanding techniques such as Wi-Fi offload and “simply” adding more spectrum, solving the mobile broadband capacity crunch and delivering data throughput over LTE networks in a cost-effective manner makes it essential to augment the macrocell network with a very large number of small cell base stations.  Accordingly, most telecom equipment manufacturers (TEM) and carriers are working to implement as quickly as possible small cell based network topologies.  In fact, recent research from In-Stat shows that network operators plan to run 12 percent of their 4G traffic over small cells by the end of 2012. As with all new network equipment, development starts at the chipset level. And the effect will be dramatic as, according to industry analysts, the small cell base station semiconductor market is expected to top $1 billion by 2016 and sales of the base stations themselves will top $14 billion in a similar period.  

In order for carriers to effectively deploy and manage such a large and distributed radio access infrastructure (with an expected deployment ratio of 12 small cells to every one traditional, or “macro” base station), base stations need to be compact, inexpensive, easily manageable, energy efficient and reliable.  In other words, carriers need “macro class” capabilities at the “micro level” (i.e., the very compact and low cost footprint of small cells). TEMs require small cell silicon solutions with unprecedented levels of scalability, performance, security and manageability with the ability to offer differentiating capabilities such as concurrent support of WLAN, QoS, interference mitigation and Self-Optimizing Networks (SON) – and, very importantly, in a form factor that reduces the bill of materials (BOM) cost now and provides a path for further cost reductions in the future.

Thankfully, a “silicon savior” has arrived in the form of a highly-integrated system on a chip (SoC) architecture.  Modern multi-core SoC architectures employ the integration of multiple DSP, RISC, and CPU cores, augmented with hardware acceleration cores, accomplishing the processing tasks associated with layers of an entire base station system’s signal and data processing. This dense integration enables base stations with ultra-compact form factors while meeting the various economic and performance objectives entailed by small cell deployment and overcoming the technical hurdles articulated below. 

A key rationale for deploying small cells is to deliver sufficient capacity closer to subscribers, particularly those currently at some distance from a macro tower.  While macro cells work well in delivering voice and 3G data services to a large area with one large cell, 4G technologies, using limited bandwidth, need cells that are smaller. Furthermore, the cost to operate and maintain, including any leasing arrangement, a macro site can reach $150,000 per month.

Deploying a mixture of small and macro cells creates what is known as a heterogeneous network, or HetNet. The HetNet creates a new radio access network (RAN) with much more complicated interference mitigation and traffic backhaul demands. The RAN devices (base stations) need to be capable of achieving error-free high throughput to all its users that demand it.  The latest SoCs feature sophisticated interference cancellation algorithms which demand high processing power. They also enable small cells to have their own backhaul capability, unlike a macro station.  This feature allows them to operate in a more efficient mesh network configuration, or, at a minimum, obviate the need for additional backhaul equipment.

Other operational elements include: processing increased number of sectors to build the highest capacity evolved packet cores (EPCs); sophisticated power management, QoS and traffic shaping tools (e.g., 2x carrier aggregation and multi-user MIMO); lower BOM costs that will allow carriers to deploy stations closer to consumers on a more economical “as needed” basis; and, often overlooked, interoperability testing with many other players in the broadband wireless ecosystem.

Multi-core communications processor manufacturers who already have experience with high-level processing requirements at the macro level have a head start and are in a position to offer critical advantages to the TEM community. These small cell SoCs would leverage proven L2-L7 multi-core technology already incorporated in Tier-1 carrier macro wireless equipment, which is very important in terms of achieving fast interoperability between RAN devices. They also can leverage this experience to create purpose-built, highly programmable baseband DSP cores and extensive 3G/4G hardware accelerators in a single chip, among other technical features. 

It also is important not to overlook the importance of software in this scenario. Important features in an integrated SoC’s comprehensive software suite package include thorough inter-operability radio conformance testing with multiple user equipment and PHY (L1) layer software that incorporates complete user-plane and control plane components for LTE, along with protocol processing libraries that take advantage of the in-built hardware acceleration. 

Carefully selecting an SoC along with L1 and L2 to L7 software and proven interoperability is essential in providing a good foundation on which a successful LTE small cell rollout can be accomplished.

Venkat Sundaresan is senior product line manager NCD Embedded Processor Group for Cavium Networks.