Snippets from field work with the University of the Philippines, summer 2017
I spent July through September 2017 working with the Village Base Tranceiver Station (VBTS) team in the UP Electrical and Electronics Engineering Department to install community-owned cellular networks in rural towns in Aurora Province, Philippines.
In Manila 24/7/2017
Manila grows wide with malls and deep with history.
Manila’s malls, one of which (the Mall of Asia) was once the largest in the world, are like mini-cities within a city but with central air conditioning to provide people with relief from the heat and humidity outside.
I live in Grass Residences right by the SM North EDSA mall. The most convenient way to get to the bus stop for the University of the Philippines in Quezon City is to walk through the mall along almost its whole length of several blocks.
SM North apparently takes its name from “Shoe Mart” which it had been many decades ago. It now consists of three five-story complexes with twelve movie theaters, two supermarkets (the Supermarket and the Hypermarket), two whole floors in the central complex dedicated to consumer electronics, an unimaginable number of other stores large and small for everything from custom rubber stamps to only mozzarella sticks, and an outdoor raised walkway densely packed with restaurants, bars, fake garden ponds with tiny bridges, and amusement rides for children. The upbeat mall jingle tries very hard to backronym “SM” to “Super Mall,” but everyone remembers the real story.
The buses I take to school are called jeepnys. These look like really tricked-out miniature school buses welded together out of sheet metal, usually open in the back and without glass in the windows. The style and name originates from the post-WWII period when these bus lines were run out of surplus jeeps left behind by American troops. When the buses eventually needed to be replaced, a family-owned manufacturer took over the task of building cars that started out looking like the original jeeps, but which buyers could order highly customized. Car bodies lengthened to hold about 15–20 people in the back; pull-cords were wired in for stops, each with its unique LED bling at the end; ceilings were padded to protect heads from bumpy roads. Now there are perhaps three different manufacturers in the Manila area, and customization is still the norm. Jeepnys don’t run on a schedule, but try to appear to fill demand. You get on the jeepny, pay the fee (saying “bayad po” for “my payment”) and wait for the back to fill up so the jeepny can leave.
According to jeepny chat yesterday (which my UP advisor Claire translated for me), some of the protestors at President Duterte’s State of the Nation address today were jeepny drivers striking against a proposed national “e-jeepny” program to mandate that they be electric.
Another of Duterte’s recent initiatives is an overwhelming police and security presence throughout malls and public places. I first found out about this from the Filipino Lyft driver who took me to the airport in Seattle. According to him, in response to a few high-profile armed mall robberies in the past few years, guards have been assigned to perform mandatory bag checks for every person at every mall entrance, and are also posted what seems like every hundred feet or so inside. Some have only batons, while a few are pointedly militarized in camo with rifles. Nearly every door has a male and a female guard. Sometimes men and women have to form separate lines behind each guard, and other times the line is single file where the male guard checks your bag while the female guard gives you the most cursory pat-down, like a gentle caress on the back. One time they asked me about a large-ish box I had containing a radio. Generally they look bored.
Claire finds the bag checks annoying enough that she avoids walking through malls. In contrast the Lyft driver had spoken positively of the “strictness” of the President’s initiative. He said the increased police presence, drug raids, and “execution by firing squad” (which he said the President favored because it was “cheaper” than having trials) made Manila safer by causing “criminals” and “drug dealers and addicts” to “stay quiet” out of fear. I asked him if he thought Duterte would ever use the strengthened police or military to eliminate political enemies; he seemed taken aback, and said no. According to Claire, some call Duterte the “Trump of the East” (“when really they should be calling Trump the ‘Duterte of the West’”). He was popular upon election with around 15 million votes, representing 38.5% of votes cast, with the runner-up having 9 million.
On the weekend I trekked out to the old city center to read the inscription on a monument to national hero Dr. Jose Rizal, a physician and writer who was executed by a Spanish firing squad in 1896 for his two novels Noli Me Tangere (Latin for “Don’t Touch Me”) and El Filibusterismo which sparked a revolution against the colonial government. They were originally written in Spanish, but are studied in Tagalog by all Filipino students, sometimes more than once (with multiple editions available at various reading levels).
I found a shelf full of them in the school textbook section of the mall’s National Bookstore— a pretty good place to find out what everybody in the country is required to know! Lucky for me, there were excellent comic book versions of both novels in English, which I am happy to recommend/lend to anyone interested. Rizal did not cut corners for fear of death in his scathing criticism of the oppressive and corrupt Spanish priesthood, ruling class, and police. Spoiler alert: nearly every sympathetic character becomes jaded or goes mad and then dies due to these corrupt individuals in power.
The inscription on Rizal’s monument is a poem, My Last Farewell (Mi último adiós), which he wrote in jail before his execution. In the original Spanish it moved me to tears — fiery and unwavering, utterly hopeless and grim yet meaning to stir hope. This I also recommend, along with other stormy revolutionary works produced under similar circumstances. I’m sure there are more than we know.
Barangay Umiray, Aurora Province, Philippines 17/8/17
So far I’ve been on my first two week-long field visits to Barangay Umiray in Aurora Province, Philippines, about a five hour drive from Manila without traffic. I’m currently on my third trip, hiding inside a bright yellow mosquito net in front of an electric fan. The technical team from UP has rented a small, bright green house behind a local council member’s for the duration of our work here. The day we rented it, he had some fresh coconuts cut down from his yard trees for us to drink and eat. :)
The house is in a sitio (roughly, neighborhood) right next to Sitio Limbok, where the first cell tower of the seven planned by UP over the next two years will be located. The seven towers will constitute a cross-disciplinary Randomized Controlled Trial (RCT) exploring the socioeconomic effects of introducing cell phone service into remote rural communities.
UP is interested in establishing a new model for public or community ownership and management of network infrastructure. The cell site and all the equipment therein, when the deployment and other socioeconomic studies are finished (within 2–3 years), will belong to the Local Government Unit (LGU) of Dingalan, the municipality in which the site is located. Through a series of Memorandums of Understanding (MOUs) facilitated by UP, the LGU has partnered with a local fishers’ cooperative, PAMANA, which will be responsible for both the day-to-day technical and business operations of the tower, and make some revenue from these operations. But PAMANA is not local enough to detect and address technical issues immediately- it’s based in Dingalan, but located in Barangay Paltic about an hour’s drive from the cell site, not within the coverage area of the tower. So PAMANA has another agreement with an almost-coop of fishers called Samahan, which is truly local to the cell tower (right next to their meeting house), but who are not used to running business operations and didn’t want to be responsible for cell tower management on their own. Samahan will be in charge of day-to-day technical maintenance of the cell site. And on top of all of this community-level organization sits our (unnamed) partner telecom, who still owns the spectrum and SIM cards the users will be buying to access the network, but will be selling them under a different branding due to the experimental nature of this project. The arrangement is complicated and there are a lot of stakeholders, but MOUs are gradually being signed and the project is moving forward.
I think this model is a significant step away from centralized telecom-based network ownership and management, and has promise. But the telecom still holds the spectrum and subsequently has the legal right to control the prices for its use, despite that prices and most other network configurations can be managed locally from the basestation. We are not in a position to change this, at least for the moment, in this context.
My research is looking into building local community capacity to maintain the cell site and recover from common failures. The site is so hard to get to that even the “local” technical university ASCOT (Aurora State College of Technology) in the city of Baler, a four hour drive away, won’t agree to help maintain it though we asked them. I am looking to discover and design ways to help people without much technical training (perhaps none at all) learn or figure out how to do the basic repairs relevant to these sites, so they can keep their cell tower running on their own as much as possible before having to call the UP team a day away in Manila.
According to Claire and Flip on the technical team, the typical way of dealing with a broken appliance is through an informal service contract with the local person who sold it to you. You’d take it back to that person and expect them to help you with it. This model works less and less well the harder it is to communicate or travel to the “point of sale,” so to speak, so sustained community capacity should be important for the network’s day-to-day functioning.
So what does a cell site actually look like, and what are the kinds of failures we should be accounting for?
[Picture omitted. Please forgive the lack of photos for now; I am posting from behind a satellite connection.]
For the past few days we’d had only the first segment of the tower mounted on the foundation and baseplate, with the antenna taped onto a 6-meter bamboo pole (from a pile of bamboo poles that we luckily found in front of our house) and propped up in the middle. Last week we had none of these things, and had to raise a fuss because the engineer we’d hired to lay the foundation had forgotten to leave some holes in the concrete that we needed for some securing pegs. (Yes, pegs- like an Ikea tower.)
As of now, the full 12-meter tower is up, and looking pretty shiny with our antenna at the top like the star on an odd Christmas tree.
We managed to create a WiFi network at the site by pulleying a smartphone up the tower (like a flag) with its hotspot capability turned on and “catching the waves” from a faraway 3G Globe network. According to Speedtest.net, the downlink rate fluctuates from around 0.3 to 3 Mbps with a pretty high latency, but this is still on average better than the VSAT link we’d originally planned at a tenth the cost and a fraction of the power consumption. We are working on replacing the smartphone with an antenna, 3G dongle, and router and use them as our “backhaul,” or connection to the wider network (the gateway in our case is hosted on some AWS servers in Japan, reached over the Internet).
This setup will sound familiar to Jessica Fujimori, a classmate of mine at MIT who deployed the same setup with me in Tanzania as an undergrad project around five years ago. Now, the cellular antenna plus stationary WiFi-capable modem is now a connectivity solution with dedicated hardware sold by Globe- there’s one in nearly every office at the LGU complex in the Dingalan town center. But we preferred the flexibility of doing it DIY and unlocked with equipment we already had.
The rest of the equipment- a cellular basestation, a single-board computer running the GSM stack, the batteries, and a network switch, are currently housed safely inside the Samahan meeting house. We’re hoping to provision a fairly waterproof cabinet to keep it all in so we don’t have to build a separate shack or mount the basestation outside on the tower.
Buildings in the provinces aren’t too well weatherproofed to maximize natural ventilation; windows have wooden bars in them rather than glass, or maybe some plastic cling-wrap-like material in a few windows. The Samahan meeting house for example is open on all four sides for better ventilation- like having windows for walls. It feels indoors in the sense of being shaded, sheltered, and furnished, but not in the sense of being sealed off from the outdoors or being designed for internal HVAC. Not that cooling isn’t important- an “electricfan” is one of the most common household appliances here, along with boat motors and televisions. But it’s important not to be dependent on electricity for cooling with the unreliable grid power.
We don’t yet have a systematic categorization of the failures that tend to happen at these sites, because the sites are relatively few, experimental, and vary widely in equipment and environment. The most common failure mode for our cell site right now seems to be power. Even without VSAT, the batteries are having a hard time charging under load, partly because the site is still crowded with vegetation which shades the solar panels sitting on the ground (they aren’t mounted yet). Luckily there is AC power in the meeting house for when the batteries really need charging, but it’s flaky with frequent brown-outs and black-outs.
Based on anecdotes of failures from other sites Kurtis has helped set up, I’ve written some lightweight scripts that help expose potentially important failure states via SMS (both push messages and query responses). Right now I’ve got a battery voltage sensor, a backhaul connectivity checker, a CPU temperature sensor (from an anecdote about equipment overheating due to overzealous weatherproofing) and a computer memory checker. I’m hoping to introduce these tools in training and see if people find them useful. I’m also hoping to simulate actual failures, such as drained batteries due to shaded solar panels, as problem-solving exercises, and have people think aloud so I can learn what they already know. Afterward the team will walk them through the solution that we’d recommend as part of their training.
I’d wanted to run this study with all 19 technical trainees before their scheduled classroom training day, but delays in getting the tower up and an eye infection* I got on my second field trip meant I only got through seven, done kind of haphazardly in a rushed group session. Running the study with them after training, however, might still prove useful in terms of predicting how skilled they’d ultimately be in a real failure scenario. Qualitatively if not quantitatively, we might be able to capture enough detail and nuance in what people do and say to distinguish old from recently acquired skills and knowledge.
I badly want my project to add real value to the deployment, but remain confused as to how to make that happen. (Real value as in helping someone, making something work better.) I have added some minor technical features. My voltage sensor is now running on an Arduino Uno that had been sitting inside the Endaga CCM1 basestation all along and used only for its 5v output to hardware-configure a power amplifier. And in the process of trying to get backhaul quickly to the site in time for my study, we discovered a VSAT-free method for backhaul. I’m trying to write deployable code, and help keep the cell site running on a day-to-day basis. But my research goals still feel a little bit out of place.
The UP team is an amazingly cohesive, charismatic group of around 10 Master’s students and recent graduates in their 20s and early 30s, who are community builders as much as they are engineers, and more of both than they are academics. They are astonishingly good at non-awkward icebreakers, and their interactions with groups of community members (even during interviews) are around 50% uproarious laughter. Many of them are also fantastic singers/musicians, and jam together in their downtime. Most of the team has been here for the past two years of the project, and all plan to continue in the lab until its completion.
Over the past few weeks they’ve shown amazing focus and dedication towards creating a sustainable deployment, from going to the field frequently to do communications legwork between organizations hours apart, to staying up late tonight finalizing and printing out all the training materials for tomorrow. Their work is much more important than mine right now for getting the system up and running, and I wish I could just help them with their goals. But I suppose the research questions I’m asking are also interesting in the long run, and the results have already yielded some insights which I’ll share in another post once they’re a little clearer. I like that UP team members translate and facilitate any activities or interviews I run, embedding their views, attitudes, and relationship with the community in those interactions.
Tonight is the second night in a row of the Fiesta in Sitio Limbok. The impressively loud music was on all day last night until 5:30 in the morning, and started again this afternoon. It’s now 3 AM and still going strong. Earlier in the evening there was more music with lyrics- everything from the Cranberries to Spanish reggae to Taylor Swift, with accompanying karaoke. Later, it has switched to something between arcade techno and really grungy club beats- think DDR with a solid “untz untz untz” in the background (sometimes also with karaoke layered over it). Giselle described it as “provincial mix,” mysteriously saying only that you’re supposed to dance to it so that your legs are really sore afterward. I wanted to go check it out but everyone else was working or sleeping so I decided to write this instead.
*** Eye infection story ***
* At first I thought it was just a bug bite on my eyelid that was causing it to swell. The local shaman heard about it through “chismis” (rumors) and paid me a visit the next day. She said there were now spirits inhabiting my eye, and it had happened (vaguely) because of a place I had been to here in Dingalan. But they didn’t mean any harm; they were just welcoming me to the region. She took a plate and drew some crosses on the front and the back in herbal oil, had me follow the the plate with my eyes as she moved it around, and burned the oil with a candle to observe the burn marks. She then anointed my head and then my eye with the oil.
The next day back in Manila, the doctor in the mall clinic sent me to the pharmacy to pick up antibiotics- much less fun. But it is a lot less red and evil-looking now. I won’t include pictures.
