Network densification with the deployment of small cells is driven by the increasing demand for mobile networks. The compact and low-power nature of small cells and a lower cost than macro towers makes them a vital component of network densification in the 5G era. However, to achieve efficient traffic supply and a good return on investment, they need to be located nearer to demand hotspots since benefits depend on the amount of traffic they carry.
Costs are linked to installation, power, backhaul, rent as well as usage. ROI comes from precision planning and careful site location near high traffic demand not well served by the existing network.
The Small Cell Forum and 5G Americas have joined forces to offer the industry recommended best practices for locating small cells within 20-40m of the ideal. The recommendations stem in part from an AT&T deployment of small cells in Manhattan for 5G with additional insights from Nokia and planning software developers Keima Wireless and iBwave.
Key Findings
Precision planning process should be based on geolocated measurements of network quality used to build up a map of how well the existing network is serving traffic demand. Small cell deployments should be in places with high traffic demand but low signal quality.
Location accuracy of the quality reports is important for the overall precision of the planning process. In one study, the averaging of observed time difference of arrival (OTDOA) network location estimates was found to have a 60m median error, limiting its applicability to small cell planning. By contrast, smarter analysis of the locate data using a machine learning approach reduced the median location error to 18m, which is within the required range.
Small cell placement is determined using an automated process that uses data and goals of coverage and dominance. The example in the report shows improved coverage and reduced costs over a manual design.
Recommended best practices
For maximum ROI, small cells should be placed as close as possible to demand peaks. Best practice is within 20-40m.
MNOs would like equipment that estimates the location of usage and quality reports to adopt smarter algorithms such as the machine learning approach demonstrated. Media locate errors less than 20m are expected for small cell planning purposes.
Machine learning models should be part of any small cell design effort. Different inputs and assumptions will be factors in the resulting models that are generated.