What challenges face radio planners when creating HetNets using a variety of new radio technologies, a mix of small and large products and a range of automated software tools? We spoke with Prof. Jie Zhang, Co-Founder of Ranplan and Chair in Wireless Systems at Sheffield University to capture his insights into the future of small cell network planning.
He discusses "outside-in" vs "inside-out" approaches, the subtle differences for planning between 3G and LTE, and how RF planning also applies to wireless backhaul.
What's the difference in approach when designing networks for LTE rather than 3G?
From a network design point of view, it's basically the same for both 3G and LTE. The planning tools hide much of the complexity. Of course, the underlying system level simulation is different for each RAT (Radio Access Technology) and there will be slightly different features for each. For example, TD-LTE also has to consider the ratio between uplink and downlink traffic.
The challenge of the next decade is how to meet demand for up to 1000x capacity increase. We can do this through
- spectral efficiency improvements (e.g. MIMO, Advanced Modulation and Coding, and Non-Orthogonal Multiple Access),
- by adding more spectrum (e.g. mmWave, licensed shared access),
- and thirdly by increasing the total number of cells.
Operators will first use the spectrally efficient FDD-LTE with newly released spectrum and many will also adopt TD-LTE in order to use all available spectrum. Worldwide, there is some 365MHz of spectrum defined for TDD technologies by 3GPP, including 8 bands for TD-LTE. So operators will have to use both in future if they want to provide high capacity mobile networks with good service.
Small cells will provide the bulk of the additional capacity, by allowing spectrum reuse many times over.
In due course, several LTE-Advanced features will also improve system performance. eICIC (enhanced Inter-Cell Interference Co-ordination) really makes the small cell work better in a full HetNet configuration, especially for urban small cells. eICIC features such as ABS (Almost Blank Subframe), autonomous HeNB power control and CRE (Cell Range Expansion) also address the interaction between macro and small cell layers in terms of interference, load balancing and mobility management. Carrier aggregation will significantly enhance data rates and help reducing interference.
Why not just use Wi-Fi?
Quality control in Wi-Fi is not as good as in LTE because fundamentally it is "best effort". When the number of users increases, the throughput will drop dramatically as the media access control in Wi-Fi is contention-based and more users mean more retries and errors and more waste of spectrum. Some people talk about Carrier Grade Wi-Fi, but it is basically a wireless LAN. Also because Wi-Fi uses license-free spectrum, interference control becomes more challenging - you don't know where other conflicting Wi-Fi access points are positioned and operating. This is much easier to manage in an Enterprise where all access points are owned by one business but you still need RF planning tools to assign optimal channels and location for best coverage and capacity and user experience.
In other scenarios, you just don't know where other APs (Access Points) may be setup and it becomes a game between them. In the future, Wi-Fi APs should be able to assess their individual radio environments and organise themselves by selecting appropriate channels and transmitting power. Cooperation is needed in order to improve overall capacity by competing Wi-Fi networks. From this aspect, sharing Wi-Fi networks between different providers will be a good choice.
What are your views on the concept of "outside-in" coverage – using outdoor small cells to cover indoor areas?
Well, I think this will depend on the situation.
For scenarios such as large shopping malls and high rising buildings, outside-in small cells will neither provide good coverage nor enough capacity. Early this year, contracted by a leading European operator, Ranplan did a study of using outdoor small cells on lamp-posts to provide indoor coverage for a large shopping area in London. In a typical outdoor 2.6GHz FDD-LTE and 2.4 GHz Wi-Fi small cell setting, it was predicted by iBuildNet® that neither LTE nor Wi-Fi could provide good coverage for indoor shopping areas in a four-floor shopping centre (the size of each floor is about 65,000 m2). It was found that over 30% of the area (1.5 m above each floor) would not have coverage.
Field measurement was carried out, which matched the prediction by iBuildNet®.
Having gained confidence with the high accuracy predicted, we did some research in another scenario with low rising traditional buildings. We published our findings in a white paper titled "Outdoor LTE Small Cell Deployment on Lampposts: A Paris City Study". One of the key findings is that outdoor 2.6GHz FDD-LTE small cells on lamp-posts will reduce the not-spots indoors to some 6% and substantially increase network and UE throughputs.
This might be a good solution for this kind of scenario where in-building installations might be problematic. However, the cost and availability of suitable sites, power and backhaul must all be taken into account.
What are your views on the concept of "inside-out" coverage – using indoor small cells to serve the streetwalks/pavements outside?
I think this could be useful for many high streets and business parks, where there are already indoor networks with high quality backhaul links. It's possible to tune the small cell antennas to leak out through windows to cover pedestrians outside. This is something that needs to be studied with different traffic profiles, something easily achieved with Ranplan's planning tool iBuildNet® which handles both indoor and outdoor environments.
Small cell indoors can be used for both indoor coverage and capacity. In some areas, the macrocells just can't provide a good service in a few specific "not spots", especially when using higher frequencies. In these scenarios, slow moving or stationary pedestrians could be handled in this way. We've found that most people typically don't move that quickly, so the system can keep up by handing off to a series of small cells along a street. Small cells are also smart enough to adjust their RF power levels to ensure a seamless footprint with adjacent cells to achieve high handover success rate.
It's important to selectively manage faster moving traffic in the wider footprints of the macro layer. Networks already detect and handle this, such as by detecting frequent changes in the reported neighbor cell lists from the smartphone. Those faster moving users, perhaps passengers in a car, are then handed over to the macrocell layer until they stop.
What about backhaul? How is RF planning related to that?
If not engineered correctly, backhaul can either become a bottleneck or an unnecessarily high cost. Fibre is preferred if it is available. For wireless backhaul, RF planning is needed to ensure the various wireless backhaul technologies can provide adequate QoS and throughput to meet demand. This is a very challenging optimisation task if you want to build a high performance, reliable and cost effective backhaul network. This is a good research topic.
Ray tracing tool is crucial to study wireless backhaul deployment. With RRPE (Ranplan Radio Propagation Engine), a fast and accurate ray-based radio propagation engine developed by Ranplan, we can evaluate within the urban areas the number of PoPs (Point of Presence) for macrocells and provide an initial evaluation of the backhaul required from the small cells – the total link bandwidths required.
Backhaul vendors can license RRPS just for backhaul planning purposes.
Disclaimer: Ranplan is a sponsor of ThinkSmallCell.com