This post describes a research study on community cellular network maintenance and repair done in collaboration with the University of the Philippines and the University of Washington.
Many small and remote communities globally do not have cell coverage. This may be because of low population densities that result in low return on investments for telecom companies, as well as challenging environmental factors such as frequent natural disasters, unreliable power infrastructure, and difficulty in getting engineers and equipment to remote areas. Community cellular networks or CCNs are full-stack cellular networks are an alternative to national telecoms and are designed to serve such communities.
The University of the Philippines’s Wireless Communications Engineering Lab is in their second year of deploying community cellular networks in a series of seven rural barangays (or towns) in Aurora Province, Philippines. These community cellular sites are designed to be simple and low cost at around $10,000, low power, minimal in terms of components, and include a solar power system if there isn’t a clean or reliable power source, which is typically the case. The deployment team is also pursuing a model of community ownership and governance for these networks, for example by a local fishing cooperative or a local government office, to promote community autonomy and sustainability.
One of the biggest challenges of running these networks is long-term, sustainable maintenance and servicing. If anything goes wrong, the communities will either have to perform repairs themselves or wait for days or weeks for an expert to come. The traditional approach is for the engineers installing the network to spend a lot of time and effort training specific individuals to perform the easier repair tasks. But that knowledge is often lost with personnel turnover and ultimately an expert will have to be on call for any more complicated situations.
Our research goal is to make these networks more reliable by increasing people’s capacity to repair their own networks without an expert. Often, repair and maintenance requirements limit the successful operation of rural infrastructure. Current best practices are centralized management, which requires travel from urban areas and is expensive, or intensively training community members, which limits scaling. We explore an alternative
model: crowdsourcing repair from the community. This study proposes that rural network infrastructure can be configured to self-report error statuses to people living nearby via automated error detection and SMS messages, which could recruit people to help. We were curious about what would happen if an arbitrary person received such a message — whether they would come to help out or recruit others, and whether they would be willing and able to attempt fixes themselves. We decided to explore this by sending recruitment messages for help with the network to the 64 subscribers present on a community cellular network in the Philippines on one early afternoon in August 2017.
Study Site: San Andres, Philippines
At this point, the University of the Philippines networks were not yet ready for launch, so we decided to start our investigation in a rural barangay called San Andres about a three hour commute over land from Manila. The network there, though it uses the same core technology as our deployments, was set up around two years ago by the national telecom Globe, and is managed very differently from ours. The people assigned by Globe to manage the network are the head of the local government, or “Barangay captain,” and the “Tanod,” analogous to a town sheriff. However, they have not been instructed on how to perform repairs themselves, but only to contact an engineer from Globe when something goes wrong.
Participation from the Barangay
After we sent the SMSs, over a third of the 64 recipients came to participate, and three quarters of them were women. We found out during the interviews that this was due in part to many men being away for work in the city during the daytime.
There was a wide range of occupations amongst the respondents, of which I will highlight just a few. We had a banana farmer, three housewives, and a recycler/trader who recruited and sent her son, a recent graduate in agricultural engineering, because she thought he’d be of help. Including the son, three respondents were all either students or recent graduates in engineering in the area. There was one woman who makes and sells bamboo brooms who described herself as a “fast learner” and seemed pretty eager to be involved. There was a real estate agent, a “lay minister” or local legal authority who handles land disputes, and a rice farmer/handyman. Basically, these were mostly ordinary people from the town skewed slightly towards having some favorable attitudes towards technology, and especially the cell network. As the real estate agent said in reference to helping with the repairs, “if we can, we would surely do it,” because it helps the whole community and “it is very hard without [the cell service].”
Barriers to Repair
We gave the respondents three sample problems where we showed them a picture of a network component that had something wrong with it. Given an SMS-length error report message, we asked them to think aloud to come to a solution.
The first alert message said that “the cell site solar panels were not generating enough power,” and the correct solution was that the solar panels were blocked by debris and needed to be cleaned. Most participants (96%) got this right, and nearly all participants gave a correct explanation of how solar panels worked by receiving energy from the sun. However, when we asked whether they would be willing to perform the repair action on their own, only 74% were willing. The others expressed discomfort, saying, “we are really not the kind of people who tamper with things if we are not allowed.” One young man even said there was a sign that said “no ID no entry” on the gate, even though when we checked, there was no such sign.
We discovered that there was a general impression of the cell site being off limits, as there was a high risk or liability if the cell site was damaged by a community member. There were also perceptions of risk of physical harm.
The second problem was that “the position of the antenna has changed,” where this is an omnidirectional broadcast antenna that provides coverage and is normally mounted vertically, but had perhaps been knocked out of place by strong winds during a typhoon. The solution was that they would “straighten” the antenna back to the vertical alignment. About half of the participants said some variant of this, while two participants gave solutions but did not state the correct solution. Only three participants felt comfortable acting on their own to solve this problem, again showing a feeling of not being authorized, but with an even lower certainty than in the solar problem of being able to do the right thing without messing up and incurring some kind of liability.
We interpreted this as a training need, and also noted that many of the people who said the correct solution mentioned experience aligning a home satellite receiver to point in the “right direction.” One of the ideas that has come out of this is, “Can we add appropriate visual cues to the equipment so that it looks more familiar, and maybe even source local parts for the tower to some extent so people are more certain of what to do?” For example, should we add a mark or even a fake reflector dish to make people more likely to point the antenna in the right direction?
The final problem we gave them was that the “CPU inside the cell site rack is overheating,” and the solution was to increase ventilation to the CPU by leaving the metal enclosure open, adding a fan, etc. Only one out of 23 solutions was correct. The lay minister said, “if it’s overheating, there’s an electric fan inside that’s not working,” even before seeing the picture for the problem, perhaps indicating some prior experience with broken cooling fans. The most common solution offered was just to “turn it off and stop using it for a while,” which, to be fair, would be a reasonable course of action. Most people referenced either their cell phone overheating while charging or that a portable TV running off a generator had overheated or even exploded due to electrical and wiring issues. However, when we asked them to find the off switch in the picture, they were unable to do so and mostly guessed based on the color. This suggests that sourcing local electrical components could help with familiarity and correct repair behavior.
To conclude, from this preliminary study we’ve learned some network design considerations that could help with sustainable rural repair with a low training cost. These include locally sourcing parts, adding visual cues to make components look more familiar or make it obvious what to do with them, and for tasks that need it, making repair instructions relatable to people’s personal experiences. As part of the work of setting up a deployment, we would recommend having a conversation with the community about making an explicit liability plan to make it clear what people are allowed to do — sort of the opposite of a clause about voiding the warranty for self-fixing.