LTE does not only deliver more capacity to mobile operators overwhelmed by traffic growth. It is also brings innovation in the RAN, where operators have started to experiment with new topologies, new ways to leverage existing equipment in new deployments, and more flexible network planning and managing tools and interference.
A substantial part of the innovation will come from Time-division (TD) LTE. For the first time, TDD (Time division duplex) and FDD (Frequency division duplexing) will have to share the show. Until now, cellular networks have used exclusively FDD with few exceptions, such as Time Division Synchronous Code Division Multiple Access (TD-SCDMA), the 3G standard used in China. With LTE, not only both TDD and FDD transmissions are included in the standard, but they are bound to coexist in the same markets, in the same networks, in the same devices.
The ability to use both TDD and FDD spectrum in LTE networks is a great opportunity to increase capacity and to tap into more affordable spectrum bands, but it is also a challenge that has started to spur innovation among mobile operators.
Softbank's new 4G network is one of the earliest examples of a TDD deployment integrated within a wide FDD 3G network. The pressure to maximize capacity, and to minimize deployment cost and time has led to a refreshing approach in network planning.
Softbank is the third mobile operator in Japan by number of subscribers, with a market share of 23 percent, and a strong focus on data services. Softbank was the first operator in Japan to offer the iPhone, and, as far as I know, the first operator worldwide to have a data ARPU higher than the voice ARPU in 2010 (Figure 1). With the traffic growth continuing unabated, Softbank is under pressure to increase capacity in its network, especially in dense urban areas--over 90 percent of Softbank traffic is generated within 26 percent of the covered footprint.
Figure 1. Voice (top) and data (bottom) ARPU at Softbank. Data APRU has overtaken voice ARPU in 2010. Source: Softbank
The new network is planned to have enough additional capacity to relieve the existing 3G network and to increase subscriber throughput. Soft-launched in November 2011, the new 4G network had 3,000 installed sites in Tokyo, Osaka and Nagoya by the end of 2011. The commercial launch is expected for February 2012, and Softbank plans to deploy 12,000 base stations to cover 92 percent of the population during the year. The network operates in 20 MHz TDD channels in the 2.5 GHz band, using the Japan-based Advanced eXtended Global Platform (AXGP) standard that is fully compatible with TD-LTE (i.e., TD-LTE devices will be able to operate within an AXGP network if they support the required frequency bands).
Interestingly, what makes the Softbank network different from other LTE networks is not intrinsically due to the fact that it uses TDD. The ability to use of beamforming is probably the major advantage of TDD, as it brings an increase in throughput (especially at the cell's edge) and interference mitigation that is not available on FDD networks.
Rather, it is Softbank assets and requirements that pushed the operator in a new direction. To deploy the network, Softbank bought Willcom's 2.5 GHz spectrum, but it also inherited its 1.9 GHz Personal Handy-phone System (PHS) network infrastructure--a very dense collection of urban sites, with 50-150 sites in a square km area, with a distance between sites that can be as low as 50 meters. The PHS network serves as the initial footprint for the 4G deployment, with sites, backhaul and antennas shared between the two networks.
From a cost and time-to-market perspective, this approach has a clear advantage. At each PHS site, four of the eight PHS antennas are re-used by the 4G network. To install a new site, the operator only needs to add the remote radio head (RRH) that can share the backhaul with the PHS network. Installation time and cost is kept to a minimum because little equipment is added at the site. No additional permitting is needed, and typically the site lease contract is unaffected. Availability of fiber for backhaul at cell sites also helped to keep costs down at Softbank, and this is an advantage that operators in other markets may not have. The base station is then housed remotely in a base band unit (BBU) hotel, where traffic and interference management across cell sites within an area is centralized. (In a typical macro installation the base station is located at the cell site), leading to better performance and lower costs.
The reduction in capex and opex is crucial for dense deployments. Of course cost considerations are also a top priority when deploying multi-sector macro-cells, but the per-site economics become much more relevant as the density of base stations increase. As we move towards smaller outdoor or indoor cells, the per-site cost have to decrease substantially to be lead to the lower per-bit costs that operators need to accommodate growing traffic loads.
The downside of re-using existing cell sites is that network planning is initially constrained to the existing locations, and coverage and capacity distribution may be well suited for PHS subscribers, but not for 4G subscribers. During subsequent phases coverage and capacity can be optimized to meet subscriber demand by placing additional small-cell sites.
But this is a price well worth paying because the fast rollout of the network gives Softbank a capacity boost in most of it footprint when it needs it, while keeping the per-bit cost manageable, and lowering the traffic pressure on the macro 3G network.
Furthermore, the decision to re-use existing cell sites forced (or allowed) Softbank to move beyond the dominating macro-first/small-cell-when needed approach that dominates LTE network deployments. Instead, Softbank uses an in-between model that does not need a macro layer for coverage (provided by 3G), but that at least initially includes cell sites which are at the high end of small cells in terms of capacity and range. Instead of using a typical multi-sector configuration, Softbank found that cell sites with omni-directional antennas, which require far less equipment and therefore further reduce the cost to install and operate a site, offer the best performance-cost tradeoffs.
It will be interesting to follow Softbank deployment and see what the learning lessons will be on performance, coverage, interference management and subscriber experience in a network with a full traffic load after the commercial launch. Other mobile operators are unlikely to duplicate the Softbank model because of differences in spectrum assets, regulation, traffic distribution, or urban topologies, but we expect that many will find equally innovative approaches that will leverage their strengths and are likely to lead to different outcomes. For instance, not many operators have a PHS network to leverage, but many do have a Wi-Fi infrastructure that can be leveraged to add small cells.
Figure 2: A four-antenna TD-LTE cell site in Tokyo that is shared with an eight-antenna PHS site (only half of the PHS antennas are used by the TD-LTE network). Source: Senza Fili