At their outset, 4G networks were defined as microcells and picocells that existed as distinct entities, serving fine-grained cell-site needs in metro and neighborhood environments. Now, the drive is to combine all cell types smaller than macrocells into a single “small cell” category. In fact, the pendulum has swung far enough to where the Femto Cell Forum, a coalition dedicated to in-building applications of digital cellular technology, decided to change its name to the Small Cell Forum.
As networks progress from current 3G to first-generation Long-Term Evolution (LTE) and then full-featured 4G, common technologies will unite the market sectors of microcells, picocells, and femtocells. Only at that point will the more inclusive “small-cell” label become justified.
Table Of Contents
- LTE, 4G, And Small Cells
- Small Cells And Carrier Ethernet Are Natural Partners
- Small Cells And Backhaul
The original drive to 4G, conducted under the auspices of the 3G Partnership Project (3GPP), was not intended merely to improve underlying speeds and transport efficiency for packetized traffic. There was also the push to unify the 3G standards that had splintered across various regions, leading to the unwieldy alphabet soup of such incompatible air interfaces as GSM, W-CDMA, and CDMA-2000.
Strictly speaking, the first generation of 3GPP’s LTE is not a 4G technology, though the follow-on LTE-Advanced (LTE-A) will meet 4G’s goals. The 3GPP defined LTE as an advanced modulation method that would take advantage of new standards such as GSM EDGE and UMTS High-Speed Packet Access (HSPA), paving the way for true 4G. In the common vernacular, many wireless operators have referred to first-generation LTE as 4G, though.
LTE is based on a unique air interface that requires its own spectrum, but offers 300-Mbit/s downlink speeds, 75-Mbit/s uplink speeds, and quality-of-service methods not available in any previous generation of digital cellular service. Its underlying architecture is based on Internet Protocol (IP), in a topology called Evolved Packet Core, and supports both time division duplexing (TDD) and frequency division duplexing (FDD). Most LTE, except in China, is FDD.
Once LTE upgrades to the new E-UTRA (Evolved Universal Terrestrial Radio Access) air interface, it will fully meet the International Telecommunications Union’s definition of 4G. By combining MIMO antenna arrays with 20-MHz total bandwidth, terminals can achieve downlink rates as great as 3 Gbits/s and uplinks up to 1.5 Gbits/s, with latencies for small IP packets of less than 5 ms.
As networks move to true 4G E-UTRA infrastructure, they become fully packet-switched. Flexibility will be significantly enhanced, with support of cell sizes ranging from tens of meters, spanning femtocells and picocells to macrocells with a 100-km radius.
Operators such as Verizon and AT&T in the U.S. and NTT DoCoMo in Japan, all of which use FDD-LTE, already are deploying the first generation of LTE globally. Smaller carriers tend to opt for TDD-LTE, since it uses half the bandwidth of the FDD version. By accelerating development of LTE-Advanced, 3GPP is hastening the rollout of E-UTRA networks, which concatenate multiple frequency bands and allow transmission from multiple cell sites to mobile users.
Adding more 4G features will elevate the importance of small cells, because the increased bandwidth and enhanced “quality of experience” (QoE) simply cannot be realized with existing cell-tower infrastructure. Operators all agree on spatial reuse of spectrum, but also recognize the crowded urban environment of macrocells and the NIMBY (“not in my backyard”) opposition to new basestations often encountered in suburban areas.
As a result, new build-out will need to rely on small-cell towers that combine low power with better aesthetics, meaning visually inconspicuous basestation equipment positioned on telephone poles, lamp posts, traffic signals, and the sides of buildings. ABI Research estimates that by 2015, there will be 5.8 million small-cell deployments, compared to less than 1 million macrocells.
Picocells and microcells are purpose-built for a packet-switched IP network. Macrocells must carry legacy 2G and 3G traffic. However, new small-cell equipment is designed specifically for IP and Ethernet, displacing legacy Layer 2 interfaces like TDM and ATM.
Small cells must be deployed as close to street level as possible, because operators cannot afford the density and high transmission power that characterizes macrocells. Small-cell sites have to be power-efficient and capable of receiving power from 120-V/220-V traffic signals or street lights. Utilizing Power Over Ethernet (PoE) standards can help keep interface power dissipation to approximately 13 W.
Precise network timing and synchronization are necessary in all modern networks based on differentiated traffic, a factor that has made Synchronous Ethernet and IEEE 1588v2 critical elements of Carrier Ethernet. This is particularly relevant in serving small-cell networks.
A traditional macrocell world requires precise timing in both frequency and phase in call handoff between towers, as well as a stable reference clock to generate radio frames for transmission between cell sites and customers. This is accomplished by receiving timing from GPS satellites. As small cells move to street level, though, maintaining line-of-sight links to a GPS satellite becomes problematic. GPS also is subject to potential jamming.
Operators in some global regions are leery of over-reliance on a satellite network owned and maintained by the U.S. government. Thus, the network itself must provide the precision timing. Requirements for LTE-Advanced are at least 10 times higher than that for TDD-LTE. Overall, adoption of LTE, small cells, and LTE-Advanced is having a dramatic impact on timing requirements (Fig. 3).
A related, but separate, requirement revolves around performance management and support for Ethernet operations, administration, maintenance, and provision (OAM&P) for both single-operator and multi-operator environments. Small cells are in less controlled environments because LTE brought a flatter and lower latency architecture to wireless infrastructure. Cell-site user and control traffic must flow back into the network, and there’s communication and coordination between cells.
Small cells introduce more networking requirements between the small cells, network aggregation points, and macro sites—hence, a greater need for OAM&P monitoring. Users have lofty expectations for uninterrupted QoE, including streaming and isochronous traffic, even across hotspots or in local sites during special events. Consequently, better differented QoS methods become necessary.
Traditional macrocell sites can exploit the rich mix of backhaul access technologies developed in both circuit-switched and packet-switched realms. Thanks to the availability of fiber interconnect, T1/E1 lines, and microwave wireless links, macrocell backhaul has become a highly contested market, with a slow but inexorable shift from analog TDM backhaul to packet-switched Ethernet over a favored physical medium.
In small cells, many of these options are unavailable due to the street-level location of the cell sites. As a result, look for microwave (MW) and millimeter-wave (MMW) backhaul to dominate in small-cell markets because many street lights and traffic lights don’t have wireline (Fig. 4).
Many deployment decisions still must be made in small-cell backhaul. Unlicensed spectrum may be considered for some backhaul. Topologies from wireline worlds, such as point-to-multipoint, point-to-point, line-of-sight (LOS), and non-line-of-sight (NLOS), all can be considered, but the physical medium likely will be a high-frequency wireless link.
The aforementioned 1588v2 network timing protocol is of special relevance to small-cell backhaul, as is OAM&P. Phase accuracy in the nanosecond range is necessary for backhaul, which can be achieved using hardware-based time-stamping along with with transparent clock (TC) and boundary clock (BC) implemented over backhaul networks.
Accurate compensation of the residence time in the TC using hardware based timestamping eliminates any dependence on network loading, internal box and queuing delays, as well as changes in (microwave) link delays as modulation formats vacillate due to weather conditions. The 1588v2 timing solutions also must be able to compensate for asymmetric link delays in hardware, since they will inevitably occur in access networks (particularly for microwave backhaul).
Improved OAM&P and network virtualization models are required as operators attempt to avoid the proliferation of small-cell infrastructure equipment. A new European model of “RAN (radio access network) sharing” may not be used in all regions among all networks, but the shared-facilities model creates an expectation of shared infrastructure, shared backhaul, and shared cells.
The shared-facilities model already has found favor in U.S. markets through third-party companies that provide common cell towers and common facilities. The viability of hybrid shared environments depends on small-cell backhaul support of multiple virtual networks and nested OAM domains. As maintenance entity group (MEG) endpoint architectures become more widely established, small-cell backhaul must support multiple MEGs.
Network managers familiar with end-to-end traffic shaping and policing seldom think about the role of traffic engineering in backhaul. Nonetheless, it’s critical from both a path-engineering and QoS perspective. Path engineering constrains the paths taken by packets and thereby minimizes latency and latency variations, which both affect LTE QoE.
Path engineering also is highly important for reliable delivery of timing over the packet network. Timing protocols such as IEEE 1588v2 rely on the upstream and downstream paths having equal delay or known delay symmetry, which could not be identified without path engineering. This is particularly important for point-to-multipoint and NLOS microwave deployments. The Metro Ethernet Forum E-LINE, E-LAN, and E-TREE concepts can be used for traffic engineering, with and without the help of multiprotocol label switching (MPLS) or multiprotocol label switching-transport file (MPLS-TP).
QoS is the other aspect of traffic engineering. Media-rich services require differentiated treatment of flows based on traffic type, customer-service level agreement, and user preferences. In the small-cell-size and high-mobility environments encountered in 4G networks, mobile users may only receive content from a small cell for 25 seconds before transitioning to the next cell. As a result, traffic management and QoS must be able to adjust much more dynamically than in legacy backhaul networks.
First-generation LTE and full 4G services may seem flatter and more predictable because they’re based on universal IP and packet switching, as well as Ethernet at Layer 2. However, the inevitable shift to small cells in both the switched network between cell sites and from cell sites to user, as well as in backhaul applications, means that infrastructure equipment must support complex wireless Ethernet topologies. The same requirements emerging for Carrier Ethernet in the wireline metropolitan environment—fine-grained OAM&P, precise network timing and synchronization, flexible support of MPLS and MPLS-TP—will be critical for small cells at user cell sites and in cell backhaul.
Today you can find very few small-cell depolyoments, other than home and enterprise femtocells. However, the movement is clear. As carriers build out their LTE networks and the demand for increased bandwidth rages on, small cells will begin to appear in dense metropolitan networks to provide coverage and capacity. All research points to increased small-cell adoption by carriers beginning in 2013 and beyond, and these small cells will require carrier-grade networking features for timing, OAM&P, and QoS.