Agent Node

This repository is the implementation of the agent node from our paper "Hierarchical Multi-Blockchain Architecture for Scalable Internet of Things Environment", which is published here. To run fully run all of the engines here, you also need to run all of the engines in the Notary Node, which is available here.

Installation

You need vagrant and virtualbox for this project. So install them first if you do not have it yet in your machine. You can download them here and here. All of the required softwares and tools has been included in the Vagrantfile and it will be installed during the vagrant up using shell provisioning scripts in ./shell directory.

To clone and setup the node.

foo@ubuntu:~$ cd ~/
foo@ubuntu:~$ git clone https://github.com/mrkazawa/proxy_node.git
foo@ubuntu:~$ cd ~/proxy_node

Note! Make sure to change the CPU and RAM information to match your needs. They will determine the CPU and RAM of your VMs. The number of CPU and RAM allowed per VM are bound to the resource in your host machine. You can change this setting in Vagrantfile.

BOX_MEMORY = "8192"
BOX_CPU = 4

To install and run the node, follow these steps.

foo@ubuntu:~$ vagrant up # if it is our first time, this will take some times
foo@ubuntu:~$ vagrant rsync-auto

foo@ubuntu:~$ vagrant ssh agent1 # open another terminal for agent1 (car owner)
foo@ubuntu:~$ vagrant ssh agent2 # open another terminal for agent2 (car renter)
foo@ubuntu:~$ vagrant ssh agent3 # open another terminal for agent3 (car backend)

Inside the SSH instances, we need to install all of the Node JS dependencies. Run this in all proxy nodes (agent1 to agent3).

vagrant@agent1:~$ cd ~/src
vagrant@agent1:~$ npm install

vagrant@agent1:~$ npm run-script # show all available NPM commands

Other useful commands,

foo@ubuntu:~$ cd ~/agent_node
foo@ubuntu:~$ vagrant reload # to restart VM
foo@ubuntu:~$ vagrant halt # to shutdwon VM
foo@ubuntu:~$ vagrant destroy -f # to completely delete VM

---

Running the Engines

Run this in all proxy nodes (agent1 to agent3).

1. Run the Storage Engine

First of all, lets create an IPFS instance.

vagrant@agent1:~$ cd ~/src
vagrant@agent1:~$ npm run ipfs-build # initiate IPFS

Get the swarm key from the boot node. Run the following command in notary1 terminal.

vagrant@notary1:~$ cat ~/.ipfs/swarm.key

This will output something like this

/key/swarm/psk/1.0.0/
/base16/
deb1aeca7acd63417d6ba13d6d8c8f894bd2d18f5dc67892214397372f562b5d

Copy the contents and create the same file in agent1 terminal.

vagrant@agent1:~$ touch ~/.ipfs/swarm.key
vagrant@agent1:~$ nano ~/.ipfs/swarm.key

# now paste the same contents from notary 1 swarm.key to this file
# Ctrl + X and then Y, to save and exit

Get the Peer ID of the IPFS boot node. Run the following command in notary1 terminal.

vagrant@notary1:~$ IPFS_PATH=~/.ipfs ipfs config show | grep "PeerID" | cut -d ":" -f2 | grep -o '".*"' | sed 's/"//g'

Copy the output, back to agent1 terminal, and replace $your_boot_node_peer_id with the output

vagrant@agent1:~$ cd ~/src
vagrant@agent1:~$ chmod +x ./storage/add_boot_ipfs.sh && ./storage/add_boot_ipfs.sh $your_boot_node_peer_id
vagrant@agent1:~$ npm run ipfs-start # to start IPFS daemon
vagrant@agent1:~$ npm test # if all is working correctly, the test should pass

Additional commands,

vagrant@agent1:~$ npm run ipfs-stop # to stop IPFS daemon
vagrant@agent1:~$ npm run ipfs-destroy # to destroy IPFS instance

2. Run the Compute Engine

At this moment, we use ganache network for our compute engine. Therefore, the agents are not running any Ethereum node. You can change the location of the ganache network by modifying the web3.js in ./src/compute/web3.js.

TODO: move the implementation to use Geth instead.

3. Run the Payment Engine

For now, we still uses one IRI and one COO. The agents are not running any IRI node. You can change the location of the IOTA network by modifying the iota.js in ./src/payment/iota.js.

TODO: move the implementation to use multiple IRI nodes.

The following snippets are steps for multiple IRIs. The IRI node can connect to one another but the snapshot is not working. Only one node has the correct IOTA balances according to the snapshot. Need further investigation!

First of all, clone and install IRI.

vagrant@agent1:~$ cd ~/src
vagrant@agent1:~$ chmod +x ./payment/1_install_iri.sh && ./payment/1_install_iri.sh

Open a terminal in the notary1 node, then run this command to get the Coordinator address.

vagrant@notary1:~$ cat ~/compass/docs/private_tangle/data/layers/layer.0.csv

Back to the agent terminal. Copy the previous output and paste it as a parameter of $your_coo_address in the command below.

vagrant@agent1:~$ cd ~/src
vagrant@agent1:~$ chmod +x ./payment/2_run_iri.sh && ./payment/2_run_iri.sh $your_coo_address

The configuration of the IRI node parameter can be tweaked in ./src/payment/config/config.json. Meanwhile, the snapshot file can be tweaked in ./src/payment/config/snapshot.txt.

---

Running the IoT Agents for Apps

This repository provides examples of IoT agents implementation for IoT apps. These agent nodes are complementary nodes that lives along with the Notary Node to provide IoT application use cases. Please refer to the paper for more details. So far, we only have a 'Rental Car' IoT application in ./src/actors/rental_car. In the future, we may add more use cases example.

Follow the instruction in here to run the Rental Car scenario.

TODO: add more IoT agents example.

---

Known Issues

If the node cannot ping to one another, perhaps it has the problem with the Avahi DNS. Try to ping to itself using the configured domain in all nodes. Then, try to ping one another.

vagrant@agent1:~$ ping agent1.local # run this in agent #1
vagrant@agent2:~$ ping agent2.local # run this in agent #2
vagrant@agent3:~$ ping agent3.local # run this in agent #3

# then try to ping one another, this should solves the issues

Authors

• Yustus Oktian - Initial work

Acknowledgments

• Hat tip to anyone whose code was used
• Fellow researchers
• Korea Government for funding this project