Are Small Cells for 5G?

5GcAs the Small Cell Summit winds down I can’t help but wonder whether small cells are a technology awaiting 5G networks. There were many discussions on architecture at the conference which cannot but lead me to ask whether the market and technology drivers are in place for mass adoption. My reading of the situation points to further delays which extends the timeline of deployment to overlap with the inception of 5G networks – whatever that turns out to be. If this happens, mass LTE small cell deployment will stall in favor of 5G. I know that I am not alone in this thinking as a few operators have privately shared the same believe stating out-right that “small cells are for 5G”. What’s said in public aside, the fact is that leading operators are looking at small cells as a part of a much larger package in network evolution that includes a rethink of the network architecture in light of evolving business models and technologies.

To put the issue into perspective, small cells were nowhere when the vision for LTE was being defined in 2004. The primary occupation then was on how to fill the largely idle 3G networks with data traffic and generate revenues for operators who sunk billions in 3G spectrum and network buildouts. SON was defined for macro cells and not for HetNets. ICIC was targeted at macro cells – eICIC and its siblings (e.g. FeICIC) were an afterthought for a “future release.” The architecture was made flat by choice and placed much of the processing at the base station. Decision making became resident at the edge which consequently made coordination more difficult. In LTE releases 8 and 9, small cells can have an adverse impact on the network. Operators who are obsessed with reliability and security struggled with a technology that introduces uncertainty. Aside from these issues, LTE networks were not loaded (and they still are not!), leaving room to grow traffic further before small cells becomes obligatory. In fact, we see most small cell deployments today are targeted for coverage and not capacity. Until the deployment driver switches to capacity, numbers will remain limited.

Mobile Subs per Region

Mobile Subs per Region [Source: Ericsson Mobility Report, June 2015]

Today the picture is different. Data growth is unquestionable, although now the issue is how to monetize data which adds financial pressure on operators who are weary of added capex and opex. Small cell technologies have evolved along a number of axes: LTE in unlicensed spectrum, Wi-Fi carrier aggregation with LTE, and split architecture with small cell PDCP layer at the macro cell (which brings back some notion of centralization). Network Function Virtualization (NFV) has made long strides and even the radio access network is now targeted for virtualization under Cloud RAN (in fact, Cloud RAN HetNets make for compelling scenarios – see here). The ideas on small cell backhaul suggest greater integration of functionality and harmonization between access and backhaul. Mobile Edge Compute is yet another emerging concept geared at enabling new services at the edge. Many technology options and architectures are awaiting the inception of a real business case that justifies the deployment of small cells. Further refinement will be required and as it will unfold over the next few years. But will we hit the 5G cycle by that time?

In looking again on the events in the last few days, we can clearly see the ‘flattening’ of the small cell market where small cells are a solution to a host of problems: IoT in buildings, MVNO models, rural deployment, aviation applications, transportation networks, etc. While all these are logical and expected applications, they are not in the traditional carrier-deployed base-station-on-a-lightpole use case. Historically, this signals a market in trouble. It can be said that the last few years were not healthy as clearly deployments were limited. Let’s hope the next few years will see a better outcome.


About Frank Rayal
Technology strategy, markets and M&As

Comments are closed.

%d bloggers like this: