Small Cells, Spectrum Sharing and Wholesale

Spectrum sharing can take a number of forms. Wi-Fi has been the classic case of shared access, as well as the classic case of unlicensed access. More recently, there is new thinking on shared access using TV White Spaces, which essentially are frequencies otherwise allocated for use by TV broadcasters, but which might actually be unused in many markets.

In India, spectrum sharing also takes the form of mobile operators agreeing to allow other carriers access to licensed frequencies, so long as reciprocity is available.  In other words, one carrier might agree to share frequencies within a single band (800 MHz, for example).

In the United States, a new form of sharing will be enabled in the 3.5 GHz band, with existing licensed users retaining primary access, but with commercial entities able to obtain licenses for secondary use, while some spectrum will also be available for unlicensed use, on a best effort basis, by devices and users, so long as primary and secondary licensees are not using the spectrum, at any place and at the same time.

Perhaps it might be said that Wi-Fi offload by mobile carriers also is a form of spectrum sharing, even when there is no formal business relationship between mobile and fixed access providers.

Small cell networks operated by ISPs generally will not be instances of spectrum sharing, though likely will provide instances of wholesale access.

Put yourself in the shoes of an executive of an Internet service provider looking at big networks of small cell access points, perhaps with per-site costs in excess of $3,000, just for equipment, and not including site rent costs.

Consider a single macrocell, where backhaul could cost $24,000 a month, supporting perhaps 160 Mbps to 500 Mbps of capacity per site. How much capacity might a small cell require? And how much could an ISP actually afford to spend on backhaul to support a small cell?

On the assumption a small cell only makes sense in a high-traffic area, the answer might be that the small cell requires as much capacity as a macrocell, in some instances. In other cases, perhaps a small cell only has to support a fraction of total macrocell bandwidth.

The issue is that traditional access using T-1 equivalents does not work. Small cells are going to require Ethernet bandwidth, between perhaps 100 Mbps and a gigabit.

As a rough rule of thumb, assume a small cell requires cost parameters about an order of magnitude less than a macrocell, with backhaul costs likewise scaled about an order of magnitude. That implies a maximum small cell backhaul cost of about $2,400 a month.

Greg Weiner, Vertix co-founder, argues that small cell sites would have to drop to about $100 to $250 per location to drive mass adoption. Again, that suggests something on the order of magnitude drop in costs is needed.

On the other hand, the cost of business grade gigabit connection is dropping dramatically, in some markets. Connections supplied by cable TV companies with dense fiber deployment (think Comcast) are one reason for thinking backhaul costs will drop to levels enabling mass small cell networks.