Gatlin IoT Farming Platform, aiming to utilize cryptonetwork utility to run algorithmic transaction to do transactions based on developing automated and efficient networks to grow food and add a tangible back to an intangible currency. Pooling resources to grow food and promote healthier and more sustainable land management.
Stack (30/May/2021)_
- Kotlin (main and embedded externals)
- Gradle w/ Groovy (main)
- Micronaut w/ http4k (maybe gRPC) (main)
- pgSQL for the whole data store (main and embedded exterals)
Main Cliffs About The Project
-
Plain truth that crypto currencies as they stand are a disaster. Insane monetary value fluctuations make crypto a bad place to invest your money. Thats is where the agricultural aspect comes in essence, a crypto network generally gains value by being able to pool resources to mine for the network. To run algorithms for to do a, b and c. So what are a,b, and c and how can I utilize this desire to collect and mine resources to my advantage? For easier conceptualization we can look at this as another avenue to buy and sell futures. Food is a renewable source that has firm and immutable timelines (this won’t be a monsanto op). I also believe that building out systems to better utilize physical resources to be something a crypto network could do. Instead of hacking me and mining my GPU’s, they can mine the field and help grow food. More on that later. Since we are still dealing with futures, we can expect do a smoother transition and offer not only quit transactions. but also off long term financial plans
-
I believe the functional style and the singleton mindset, coroutines/immutability constraints and general design of Kotlin perfect for not only building out machine learning services to advance food science, but I also believe it to be the perfect language to use in a “bastion” server for the fog model I have planned out in terms of handling resource management of peripheral devices. I don’t see Kotlin now as a place to take the embedded stuff (that would be cool), but it acconels everywhere else I need it and its just fun to use.
-
The project will be using Kotlin for the embedded part of the embedded land management system. New development as of 30/May/2021 as I just found out the native platform can run on embedded system without the JVM being needed as well as it having interoperability with C. This is great news as it eliminates another upkeep issue with multiple technologies.
The embedded systems will run various tasks determined by the Bastion Scheduler “EtherAgent” which then designates it a task depending on transaction requirements. With the C capabilities, I feel better equipped to handle the low level tasks needed to keep a lean peripheral and minimize resource expenditure issues with energy saving requirements needed to make the usage of peripheral devices a viable option to do long term work with a single battery charge. -
The company is going to get started in two places to do an alpha prototype and an alpha deployment. First step is to utilize current tech to start simulating communication interfaces using home automation tools to establish a top down ability to communicate. From there I plan to get started by using grow operations to move forward and develop the laser reading mechanisms, the image processing mechanisms and build out scalability to be designed around the ability to operate in potato fields, wheat fields, corn fields. To have this protect actually work properly the network will have to extend into a variety of food markets to gain any adoption to it or at least its concept centered around sustainability. The basic idea is to use kotlin to manage a bunch of mini node servers running land management operations to produce value.
-
NLP based security system based around cochlear mnemonics and vernacular. This is to train and get off the ground. I chose this path because of the transient but effective uses of language. I basically want to build out a bunch of chatbots that speak swahili backwards, real wind talkers to handle encryption and decryption. I believe that language and the ability to get our AI and Neural Networks to be able to develop its own kind of “vernacular” to be a very good step forward for cyber security. If we can handle potential threats at the input box, I think we should. I have some ideas about how to go about this, mostly utilizing the flowy and no blocky good stuff in javascript. I’m benching the development of this for now as I am not confident in making a decision or starting that development work up
-
PgSQL can handle the entire datastore. It can handle the functional db’s the ints, the prods, the app db’sm, the nosql with its json and bson capabilities. Its 100% open source and it is a tried and true platform. I know theres a lotta push for “graph” databases, I don’t know the benefits and I won’t pretend to. However, I believe pgsql would do just fine with data warehousing and machine fine tuning with a good set of star tables I believe SQL to be more than capable of handling these types of relational requirements. I’m using the json/bson functionality to store configuration files with are the only bilateral part of the current UDM concepts. From what I read it handles itself very very well. Plus, I never hear about the pg team in the news. But I don’t. talk to people much, so I may be wrong.
-
I plan on using Docker Compose for the container management. There are two reasons behind this: I do not like the idea of the HUUUUUGE yaml file I’d have to write for a K8 Cluster implementation and docker compose allows for runtime configurability which is a huge plus for me considering the current conceptualization of how the pieces work together and how the UDM and configuration setting work. I could be wrong, again, I’m really out of depth in the undertaking.
-
It will be an Agent Model System. This is the right way to go for architectural decisions and especially in facates like IoT/Embedded Systems. The closer we go to the edge, the more compartmentalized out peripheral and mesh systems need to become. In a sense, almost independent. For that reason I have a hierarchical structure resembling the military and agents are akin to military grades in terms of their scope, functionality and necessity. With the current UDM I am utilizing a “UpChain” and “DownChain” methodology. Commands only go one way, one route. Either up, or down. Now to distribute commands to such a system I am incorporating the “Agent” type to be the bilateral messenger. In a sense I want to build the deployment services out to run their own configuration schemes based on policy changes coming from upchain agents who then send orders down chain to the peripherals to do work.
-
All communication is going to be routed through the sidecar. the sidecar will handle things like state management, audit logging and general system necessities like heart-beat checks. With that I also need to incorporate a GNU Radio like service to handle EM wave sanitizing. This is a place I would really like to use Kotlin to do the heavy lifting. I believe Kotlin to be a great language to do and handle various signal processing services.
-
I need guidance on what to use for the machine learning parts of this project. Earlier drafts focused on BERT/TensorflowTFX/Flink Flip-39 Runner to handle AI. I also had an ONNX draft. I would prefer to stick to my guns and go with my gut: Kotlin. I’m just unsure of what that looks like in present.
I appreciate the Kotlin member for allowing me to post this hear. It’s been a good help to get this out on paper.
Thanks!
Alex - Update 30/May/2021 I