Summary

  • Heimerl’s research moves from a locally operated GSM network in Papua to a multitenant system that partnered community networks with a Philippine mobile operator. In both, local service depended on an operating boundary that mattered as much as the radio equipment.
  • One profitable site and a multi-year deployment show what the model could do in particular settings. They do not establish universal profitability, complete local control, or authorization by every resident affected.

A network is local only where its controls are local

When a mobile operator cannot justify a tower and backhaul for a small, remote market, the choice is often framed as either wait for national coverage or accept no service. Heimerl’s community-cellular research proposed a third operating model: use small sites, existing local capacity and local operators to make a cellular service viable at a scale that a national carrier may not pursue.

The key claim is not that a community can reproduce every function of a national carrier. It is that some decisions can move closer to users. The difference matters. “Community-owned” can describe the site or service without describing who holds spectrum rights, issues globally reachable numbers, approves interconnection, sets shared policy or sees network-wide records. A useful account has to locate each of those controls separately.

Heimerl’s 2013 Berkeley dissertation documented an early experiment in Desa, a rural community in Papua, Indonesia, of about 1,500 people outside existing network coverage. It reported more than 100,000 communications during six months, a 56.6% reduction in night-time power draw and more than US$5,000 in revenue. The dissertation estimated more than US$368 in monthly profit for the local operator even when the installation was assumed to be fully financed.

Those results made the local economics legible: an operator could combine a small footprint, locally useful services and lower power needs rather than wait for a large carrier's standard business case. The figures remain a single deployment's reported outcome. They do not tell an investor what a new site will cost in another country, how unpaid labor should be valued, or whether the same demand and power conditions will recur.

CCM made the carrier boundary part of the design

The later CommunityCellularManager (CCM) work addressed a different scale problem. A local network could shape service, but a path to the wider telephone system required scarce resources and integration that small operators generally did not control. In the Philippines, Heimerl and colleagues designed CCM as a shared controller for multiple community networks working with Globe Telecom.

The arrangement split authority by function. Local networks could offer distinct services and continue some communications when backhaul failed. Local operators handled site-level service and billing. Globe supplied licensed spectrum, phone-number blocks and interconnection, and retained controls over spectrum use and parts of subscriber provisioning, administration and global network policy. CCM gave the carrier one point to manage the relationship with many small sites while letting local networks keep room to adapt.

As of December 2018, the USENIX paper reports about 2,800 monthly active users across 17 launched sites. It also says the first integration with Globe took just over four months and required approvals and configuration changes across several carrier teams. The scale was real; so was the cost of making separate organizations and systems work together. The paper describes a partnership, not a transfer of the carrier’s regulatory or interconnection position to each community.

That distinction is the central operating lesson. Local autonomy can be engineered inside a shared system, but its boundary is set by interfaces, permissions and scarce inputs. In CCM, local service did not require the carrier to expose its whole core network to every village operator. The controller mediated access and made the partnership administrable. The same arrangement also left the carrier with a privileged vantage point and the ability to define what each network could do.

Usage changes the business case

A separate 2021 study by Heimerl and colleagues examined a data-only community LTE network in Bokondini, in Papua’s highlands. It should not be combined with the earlier Desa GSM experiment or the Philippine CCM deployment. The researchers analyzed one year of usage and financial records from a different operator and site.

The study found irregular demand, not continuous mass use. Forty-five percent of users were offline more days than online. The median user consumed about 36 MB per day averaged across the study period and about 77 MB on days when they were online. People often bought small data amounts; a few heavier users supplied a large share of revenue. Most users spent less than US$1 a day, yet the paper reports that revenue exceeded the measured costs and the network remained financially sustainable during that period.

The cost structure explains why those qualifications matter. Reusing local infrastructure and local installation labor kept reported capital cost below Rp150 million, about US$10,000. The study lists roughly US$95 per month for maintenance and US$300 for satellite backhaul. It also notes that informal user support was not included in the cost accounting and that a few top users contributed substantially. This is evidence that a small network can work under a particular combination of local assets, labor, demand and backhaul costs—not a general margin forecast.

Irregular use can be a service design choice as much as a demand problem. The study describes residents using small top-ups and sometimes relying on a national carrier’s 2G service for urgent messages. In that setting, an LTE data connection was not an always-available substitute for every other network. It supported a narrower set of tasks with its own cost and availability profile.

Participation is evidence, not a denominator

Heimerl’s co-authored work on community network management also exposes a question that performance figures cannot settle: who gets to shape rules for a shared service? A 2021 study in Santa Inés, Oaxaca, explored possible congestion-management tools through workshops and interviews. Participants raised both collective needs and individual privacy, including how much user activity an operator should record and who could act on the resulting data.

The study is a design inquiry, not a deployed policy or a vote by the whole community. It involved one site and 33 participants, skewed toward men; only some participants had direct experience with the LTE network. The authors also note that local authorities organized workshop attendance and that existing relationships could have influenced which topics people raised. Those details do not invalidate the findings. They define what the findings can represent.

This is where the word “community” needs an operating definition. A network may be locally installed, managed by nearby operators and discussed in open workshops, while its users still differ in access, experience and ability to influence decisions. Participation gives operators knowledge and can reveal preferences. It does not, on its own, prove that the people who attended authorized every rule or represent everyone who bears the service’s costs and effects.

Heimerl’s lasting contribution is therefore not a universal recipe for self-sufficient telecoms. It is a body of work that makes local adaptation technically plausible and then shows the dependencies that remain: carrier resources, backhaul, power, maintenance, concentrated demand and the quality of local decision processes. For operators and policymakers, the question is not whether a network is “community” or “carrier” owned in the abstract. It is who controls each input, who can change each rule, who bears each recurring cost, and what evidence supports the claim that the service works for its intended users.

Sources