Thursday, August 6, 2015

Wireless Operational Link for the Field (WOLF) - Results

This is the final post of a series of posts dedicated to my response to challenges deploying wireless access to research plots at Toolik field station.

Results

Overall, the entire experiment went very well. Wireless network was delivered to roughly 60% more research plots than prior to deploying the WOLF rig; for a price of around $700. An additional setup deployed on the other side of the hill would have increased coverage to approximately 80% coverage. Major issues encountered are listed below:

Issue #1: Basic Network Control: The biggest hurdle when deploying an AP that doesn't involve interaction with the IT network administrators is ability to extend existing IP ranges. This would have been desirable for managing remote sensors via IP. As a work around, I was eventually able to find a web-based service to manage my Internet of Things (IoT) Intel Edison devices communicating over http ports. Being a former HPC systems administrator, this was a difficult adjustment for me from IP to http port control as I am used to being able to ssh to my systems. However, the meshcentral.com service gave me 95% of what I needed to manage the remote sensor boxes.

Issue #2: Battery Longevity: My tests at home with mild traffic resulted in the Goal Zero battery lasting about 12-15 hours with both AP and high-gain antenna. In the field, the battery lasted for a much shorter time, roughly 9-11 hours. Even when a second solar panel was added, the performance under moderate traffic was still only slightly better. A possible reason for shorter battery life may be due to the AP occasionally losing the Toolik wifi signal. It was also overcast for a few days in a row. Also interesting is I found the Toolik network itself occasionally drops packets even with good signal. I don't believe weather (cold specifically) was a factor as the temperatures were very warm during the deployment tests.

Issue #3: Geophysical Limitations: The hills around Toolik are solid rock. Wifi signals don't go through solid rock well if at all. This reality made the wifi range uneven. In addition, the rocky hills also made the antenna pattern very noticeable. For instance, the high gain antenna emits wifi signal in an omnidirectional pattern. I found that if I stood in a lower valley, the signal from my WOLF unit disappeared even though I was physically within range. Similarly the Toolik wifi signal extended further into the field at the lower elevation, which happened to be at a similar elevation. Again, at a higher location on the hill, the same Toolik wireless signal could not be picked up at much closer range than in lower areas.

Final Remarks

The WOLF network experiment is a first step towards delivering network connectivity to research plots in the field around Toolik Field Station. The equipment was chosen for ease of use, portability and cost over high performance. The configuration of the WOLF setup is not intended as a long-term, permanent solution for wireless extension at Toolik. Instead, this experiment presents an alternative solution that can be deployed today with little effort from Toolik operations. This component of the Arctic Glass project is an initial survey to examine existing wireless range, as well as a "quick and dirty" interim solution until long-term solutions are put in place. All of which is in support of other experiments for the Arctic Glass project; including remote sensor connectivity to cloud storage and Google Glass retrieval of data while on the WOLF extended network.

Wireless Operations Link for the Field (WOLF) - Survey and Design

This post will discuss the design I put together for a portable, solar-powered, easy-to-use wireless solution for Toolik research plots located outside of the camp's immediate wireless range.

First I'll summarize deployment procedure:

PROCEDURE: Overall, the WOLF rig went in pretty smoothly. We did a little bit of engineering hacking to find a good stand (surveyors tripod), grounding method (wire and rebar), and extra insurance against the elements (battery went partially in a plastic bag). We also did a rough wifi signal survey to find the edge of the existing Toolik wifi network to figure out the best location for the rig. The prep done with the router/AP itself was minimal. I changed the admin password, connected to the existing Toolik wifi in AP mode, and upgraded the firmware. I didn't bother with security on the wifi as the Toolik network is open.

Design
 The WOLF rig was intentionally designed to be a "bare minimum" outfit that any researcher, regardless of networking expertise, could put together. There was also significant consideration given to the portability of the rig as some researchers expressed interest in temporary wifi setups that could be brought into the field and taken down the same day. Components of the setup that were easy to purchase and intended for consumer-level experience were preferred over components that required more technological expertise to work with. Finally, reasonable cost for performance was a significant factor in component choices. **Note: 2.4GHz was chosen intentionally for 2 reasons: 1 - backwards compatibility if the Toolik wifi hadn't yet been upgraded to 5Ghz. 2 - Physically limit bandwidth consumption from the WOLF network into the Toolik network to prevent disruptions of the camp's wifi service.
** Cooler was not used given warm temperatures **

Cost Breakdown

Router/AP: $214
High Gain Antenna: $120
Battery / Generator: $230
Solar Panel: $89
Waterproof Ethernet, Antenna, Adapter Cables: $58

(does not include stand or grounding supplies)

Total Setup Cost: $711

Specifications

The following are specifications for each of the major components of the WOLF rig.

-- Router/AP --

Model: Hawking Technology HOW2R1

Network Specifications
  • IEEE 802.11b/g/n
Frequency
  • 2.4000~2.4835GHz
Speed
  • Up to 300Mbps
Interface
  • Wireless 802.11 b/g/n or Ethernet 10/100M
Power
  • 48V, 0.5A Switching Power Adapter (PoE)
Security
  • WPA, WPA2, WEP
  • Full Router Security Features: Firewall, Demilitarized Zone (DMZ), URL Blocking, Network Address Translation, MAC Address Filtering, IP Address Filtering, Access Control, Denial of Service (DoS)
Receiving Radio
  • Antenna: 1 x 11dBi Directional Antenna
  • Output Power:
    • 11n:18 ±1.5dBm
    • 11g:21±1.5dBm
    • 11b:23±1.5dBm
Broadcasting Radio
  • Antenna: 2 x External 5dBi Omni-Directional Antennas
  • Output Power:
    • 11n: 15±1dBm
    • 11g: 15±1dBm
    • 11b: 18±1dBm
  • Broadband Router Features: DHCP Server, Firewall,
    Demilitarized Zone (DMZ), URL Blocking, QoS (Quality of
    Service), Virtual Servers/Port Forwarding, Network Address
    Translation, MAC Address Filtering, IP Address Filtering, Access
    Control, Denial of Service (DoS), Remote Management
Hardware
  • Repeater:
    •  Antenna Jacks: Two N-Type Connetors
    • One 10/100M Ethernet Ports
  • PoE Injector:
    • Power + Data Out Port
    • DC 48V
    • LAN Port
Product Weight
  • Repeater: 481g or 1.06 lb
  • Antenna: 68g or .15lb (each Onmi-Directional Antenna)
  • PoE Injector: 6g or 0.1 lb
Product Dimension
  • Repeater:6.9(H) x 5.5 (W)x 3.5 (D) in
  • Antenna: 8(H) x 5.25(W) , Base: 3.5(D) in
  • PoE Injector: 3(L)x 2.7 (W)x .75(H)in
Temperature
  • Operating: -20~70°
  • Storage: -40~80°C
Humidity
  • Operating: 10~90% (Noncondensing)
  • Storage: 5~95% (Noncondensing)

Manufacturer Website: http://hawkingtech.com/products/hi-gain_wireless_networking/high_power_wi-fi_solutions/how2r1.html

-- High Gain Antenna --

Model: Hawking Technology HAO15SIP

Network Specifications:

  •  IIEEE 802.11b/g
Electrical Properties
  •  Frequency: 2.4~2.4835 Ghz
  •  Impedence: 50 Ohms Nominal
  •  Gain: 15dBi
  •  Radiation: Omni
  •  Polorization: Vertical
  •  H-Plan:360º
  •  E-Plan: 30º

Manufacturer Website:http://hawkingtech.com/products/hawking_products/outdoor_wireless_solutions/hao15sip.html

-- Solar Battery + Panel --

Model: Goal Zero Yeti 150 Solar Generator / Battery

  • Cell Type: AGM Lead-Acid
  • Peak Capacity: 168Wh (12V, 14Ah)
  • Lifecycles: hundreds of cycles
  • Shelf-life: Keep plugged in, or charge every 3-6 months
  • Fuses: 20A, user replaceable fuse
  • Management system Charging and low-battery protection built-in
  • USB port (output): 5V, up to 2.1A (10W max), regulated
  • 6mm port (output, 6mm, green, hexagon): 12V, up to 10A (120W max), regulated
  • 12V car port (output): 12V, up to 10A (120W max)
  • AC inverter US (output, 60Hz, modified sine wave): 110V, 0.7A (80W continuous, 160W surge max)
  • charging port (input, 8mm, blue, circle):14-29V, up to 5A (60W max)
  • Operating usage temp.: 32-104 F (0-40 C)



Model: Boulder 15 Solar Panel

  • USB Port: 5V, up to 1A (5W max) regulated
  • Solar port (blue, 8mm): 14-22V, up to 1A (15W max)
  • Rated Power: 15W
  • Open Circuit Voltage: 18-22V
  • Cell Type: Monocrystalline

Manufacturer Website:
http://www.goalzero.com/p/164/goal-zero-yeti-150-solar-generator
http://www.goalzero.com/p/20/boulder-15-solar-panel


Deployed!

(Tripod is a repurposed surveyor tripod.)



Next post on this subject will discuss the results of this experiment....


Tuesday, August 4, 2015

Wireless Operational Link for the Field (WOLF) - A Need for Wifi in the Field

In this post I will discuss the rationale, architecture, and results for developing a portable wireless rig for scientists conducting fieldwork. I should first mention the need for wireless connectivity to research plots is not a new concept. At the Toolik Research Station, "wireless for the final mile" has been almost a holy grail wish list item. With wireless connectivity researchers could deploy sensors to monitor their work, but also could use the internet connection to find reference documentation while working in their plot. The realization that all equipment as well as equipment manuals to research references must be prepped for offline use is almost unthinkable to a smartphone society that takes constant connectivity for granted. Yet even while the need for internet connectivity is justified, the cost of surveying, designing, deploying and maintaining a permanent solution has been prohibitive in both personnel availability and cost.

Ok. I will back up a bit now to explain a few things not necessarily apparent to those not directly involved with fieldwork....To fully understand what a "research plot" is, I will demonstrate here:

For research that involves the monitoring of vegetation cycles, gas release (such as co2) from the soil, and other measurements of interest; researchers set up experiments on a section of ground called a "research plot." The plots must be setup in coordination with the station's GIS staff in order to comply with Federal permit regulation of protected land issues as well as preventing one scientist to set up shop on top of another scientist's experiment. For reference, the map below shows the established research plots located near the Toolik field station. Notice the concentration of plots south of the lake. This is "the next mile", which also happens to be located over a reasonably-sized hill comprised of solid rock (not wireless signal friendly).



If you are interested in the comprehensive overview of the Toolik plots, please refer to the GIS informational page for plots/permits.

An actual research plot up close looks something like this:


This plot in particular, utilizes a greenhouse. The location of this plot is in the middle of the mass of yellow dots south of Toolik Lake.

At this point, it should be apparent why wireless access to research plots in an area as remote as the Toolik station is desirable but presents challenges. It is also worth mentioning that wireless to "the last mile" is a fairly easy task compared to delivering internet connectivity to plots located 20 miles away from camp. Which, also happen to be accessible only by small helicopter.

Next post on this topic will discuss the design I came up with for the WOLF rig...



Time to play blog catch up!

The Arctic Glass project has been moving along quite well, and I finally am able to sit down and document via this blog. Not only has the project weathered the storm that is Google's removal of Google Glass from the public market, but also now includes an Internet of Things (IoT) component. The IoT inclusion was to demonstrate capabilities of real-time data monitoring and collection using only consumer-grade commodity hardware (Intel Edison microcontrollers + Google Glass + Freeboard dashboard). My second visit to the Toolik research station in Alaska has also yielded new insight applied to real-time data monitoring in remote locations. I will be adding the various posts with more detail on these topics during the next week or so. For now, here is a pic of the portable wireless network I designed. The configuration is minimal and does not require networking IT assistance, so virtually any researcher could easily setup for themselves. The connection to the station's wireless backbone is through AP mode only, so no additional IT skills are need beyond what would be required to setup a basic wireless home router. The rig is solar-powered and with a -15dB high gain consumer-grade antenna, was able to extend wireless signal to the majority of research plots at the station! A second wireless rig could have picked up a few more on the other side of the hill (solid rock make wireless signal very angry). You can see the field station in the lower left corner of the picture. More details to come on the rationale, architecture, and results of this experiment and others very soon.....



Sunday, November 30, 2014

Citizen Science Data is the New Data Bitcoin

The Arctic Glass project has spiderwebbed out into a handful of directions I hadn't originally planned, one being "how to capitalize on citizen science." Originally citizen science came up only briefly on my Toolik visit, and was specific to solving instrument validation issues for USGS stream gauges in remote locations. I spoke for a while with the techs doing readings in hard-to-reach places like Barrow, AK, or even at the non-helicopter only sites like a truck ride a few miles up the pipeline haul road. Getting to these locations are time-consuming, and a misbehaving gauge doesn't often show up as an issue until the data is missing. What if there was "a guy" or someone local to the area that goes by that particular gauge on a regular basis. This "guy" might not have the training to fix instrument issues, but what if there was an early warning light or something simple to tell an average observer the instrument is having issues?

And so has been the general extent of citizen science to date. Scientists set up the observation, citizens help with the readings. This has been a great success on websites such as Zooniverse.org, and has even resulted in new discoveries coming directly from observant enthusiasts. But what happens when the volume of data we are churning out on a daily basis in our cities, airplanes, ships, and even personal devices eclipses the volume of data collected by the traditional scientific process?

This is where I personally find things to get really interesting. Here is the concept I have been playing with: What if data wasn't a product, but a currency? In that I mean, has transactional value that is measured rather than volume for volume's sake? In this model, we might view scientific data as similar to traditional dollars. Reasonably well structured and monitored (with those pesky data management plans) at the least. I akin traditional data collected by scientists and mostly for scientists just like the good 'ol dollar bill. If we assume traditional scientific data is like the dollar: trusted, backed by known "value", regulated....where does that leave data collected by citizens? OR, citizens that are also scientists, but not working in their given domain for a particular project, just for fun? (In other words, we can't assume the data is poorly collected and therefore less valid.)

So then, it looks to me like citizen science data is to data as bitcoin is to dollars. It's untrusted, wild, making its own rules, with everyone trying to constantly validate it on an old standard for traditional data that may no longer apply. Most importantly,  in the future citizen science data will most likely be more plentiful and more granular (think local collections in houses next to each other) than traditionally collected data. It's even feasible a market for citizen science data could flourish along side the market for harder-to-get scientific satellite data.

Just like bitcoin, only time will tell I suppose, if this prediction becomes a reality. However, I am feeling fairly confident here given the technology to perform credible scientific work is available. The first thing we need to figure out is value of citizen data, and also traditional scientific data for that matter. For example, at this point we aren't even sure if our data is worth a chicken, a pig, or two bags of flour....or if value is variable depending on transaction participants.

Friday, November 28, 2014

Promoting Awareness and Attracting Collaborators: American Geophysical Union Conference 2013

As the 2014 AGU conference quickly approaches, I'm taking some time to document my time spent at the American Geophysical Union conference in December 2013 in search of collaborators for the Arctic Glass project. I gave a short presentation during the polar cyberinfrustructure track at the conference. From this and other discussions held on the conference trade floor, I gathered approximately 20 contacts that were 'definitely interested' in exploring the use of Google Glass with their work. Of those 20, 7-8 contacts had definitive ideas regarding what projects they would pursue using. This is even more of an interesting metric to me now we (Scientific Computing team at AWS) are experimenting with a similar approach to finding collaborators this year at AGU. I am also very interested to find out if the Arduino-style "internet of things" will be more prevalent this year. There's a lot of promise in Maker devices to measure in-home toxins, back yard garden soil richness, and even your own biometric activities. The internet of things will rely on the rapid prototyping abilities of these types of DIY boards, which in turn means the probability federal agencies involved with local scale environmental modeling enlisting this type of sensor data collection is very high. All of this for another post for now...

Overall, it definitely seems targeting the appropriate professional event for the community you wish to engage is still a tried and true method of making collaboration connections. I am curious to see how the upcoming fall 2014 meeting pans out for AWS, because the connection I made at AGU 2013 that enabled this project was at the Toolik Field Station exhibition floor booth.


Tuesday, September 2, 2014

Experiment Design: Facilitating Awareness to Rapid Development

In general, there is a significant gap between consumer technology development and technology intended for scientific use. However, the two seem to be on an exponentially accelerated path to collision and have been for some time. For example, the core development of smart phones developed almost completely outside the concept of designing instruments for science. It's true that research specific to cellular transmission, graphics acceleration, hardware improvement, ect, can often be attributed to academic research...however, the product itself, the smart phone, is entirely designed for general consumer / commodity use. It was only after mass use did scientists start to realize the potential of the mobile sensors, data collection, visualization, and other uses specific to the need of experiments. At the same time, the pure science development being released for mass consumption (LDAP protocol is a good example) in a final form has reduced over the years. This disconnect, which came about organically from the pace of technological invention, has left a large void in which science could be capitalizing and even leading the development of, commodity devices both the general population and science could benefit.


  • So then, how to better integrate new, perhaps revolutionary technology in particular, to the scientific workflow? 
  • On top of that, how can scientific study influence commodity devices such that the devices are primed to work as citizen science or teaching and learning devices once development is mature?
  • Finally, how can commodity devices, both cheaper and readily available in support and quantity, both accelerate discovery and improve cost effectiveness in hard to reach regions such as the Arctic, Antarctic, Ocean, Space and others?
  • What potential advancement in science methodology and workflow could result from newfound researcher awareness?

These questions led to a collaborative approach between myself and the National Science Foundation Division of Polar Programs. The technology of focus for the experiment is a result of a potential data collection and validation method created from new disruptive devices and services.  For visualization and collection, I proposed a revolutionary device still in early stages of development be a centerpiece of the project: Google Glass. Additionally, Amazon Web Services would provide the computational power and scale necessary to address in-situ processing for data collection and retrieval. Finally, additional resources yet to be named would provide for additional sensor immersion and high-accuracy data collection in tandem with the Glass and AWS technologies.

With Polar Programs, we discussed many strategies, but ultimately settled on a somewhat hybrid approach to soliciting projects. The traditional aspect of our plan focused awareness efforts on well-know gatherings related to earth science domains we wished to target. the American Geophysical Union (AGU) conference was unanimously determined a good fit for a "wide net" approach for soliciting interest in pilot projects. Once there was specific interest that could be tied to specific scientific discovery, the AGU opportunities will evolve into targeted projects delivering proof-of-concept examples. Finally, the proof-of-concept projects are to provide a basis for extrapolation of additional areas of scientific discovery to be pursued.