Skip to content

Latest commit

 

History

History
150 lines (125 loc) · 6.02 KB

old-README.md

File metadata and controls

150 lines (125 loc) · 6.02 KB

TODO 13th November 2023

DEVICE_INTERFACE

{
  "key_name": <hash of key name>,
  "host": <host>,
  "port": <port>,
}
  • master server
    • listen on 33000
    • GET "/pause?key_name="
    • POST JSON "/register" 
      • {
  "emulator_interface": {
    "host": "10.35.70.37",
    "port": 34000
  },
  "devices": [
    {
      "key_name": <hash of public key>,
      "public_key": <public key PEM format>, 
      "host": "10.35.70.37",
      "port": 33001,
    }
  ]
}
  • slave server (naarora)
    • listen on 34000
    • python slave_server.py --master-host 10.35.70.35 (default --master-host to 127.0.01)
    • on startup make "/register" request to master
    • needs a "/update_topology" endpoint
    • POST JSON" "/update_topology" 
        [{
"key_name": <hash of pk>,
"host": <host>,
"port": <port>,
"neighbours": [DEVICE_INTERFACE, DEVICE_INTERFACE]
}]

TODO

  • design devices, enumerate sensors and actuators
  • design the "highly disconnected" scenario and use case to test/evaluate against
  • improve performance
  • improve code
  • handshake to swap identities
  • preshared trusted list of identities
  • metrics and visualization
  • run on the pi
  • interop with Ted's tcdicn

ICN Emulation

We name our protocol g17icn.

This document should describe g17icn version 0.1.

CS7NS1 Project 3 asks us to conceive, design, implement, build and prove a robust, secure, sufficient peer-to-peer networking protocol based (mandatorily!) on ICN principles.

We are also meant to conceive of a highly disconnected scenario in which the network runs and emulate various devices.

Scenario: Under water networks for climate and biodiversity monitoring and modelling.

Motivation/Application: Ocean temperatures, currents, El Niño and La Niña phenomena, are hard to predict and have huge implications for the weather experienced around the world. The increased number of rapidly escalating oceanic tornadoes is attributed to rising ocean temperatures; The New York times reported on "The Daily" an example of rapid escalation on 27th Oct. 2023, noting "Hurricane Otis transformed from a tropical storm to a deadly Category 5 hurricane in a day, defying forecasts."

Well, to have a hurricane, you need to have warm water. It has to be 80 degrees Fahrenheit or higher to really give it the energy that it needs. So think of a hurricane as like an engine. and that energy, that warm water, is the fuel that’s fueling the hurricane. — Judson Jones

Having more realtime data about oceanic temperatures, not only at ocean surfaces but also in the depths, would help meteoroligists such as Judson Jones make more accurate predictions about the escalation of storms. Early warning and accurate prediction can help give vulnerable communities enough time to evacuate, saving lives.

Notes on g17icn-0.1:

  • The protocol is built on HTTP and JWTs. Devices run HTTP servers that listen for connections and process both interest and satisfy packets.
  • In version 0.1 we use JSON Web Signatures (JWS) and not JSON Web Encryption. See the JWT RFC 7519.
  • Packets are encoded as JWTs, both interest and satisfy packets, but some additional meta data about the interest or satisfy packet can be included in HTTP headers.
  • All packets are signed by the producer of the packet. In version 0.1 we are using the RS256 algorithm to sign the JWTs. This prevents tampering, but does encrypt the data or keep it private.
  • The protocol is lazy, in that data is only sent to the network after positive confirmation that there is interest in the data.
  • TODO(optimization): batching. Since a packet is a self-contained JWT/JWS, we can actually send multiple packets in a single HTTP request. We should investigate whether this can be leveraged for performance gains or reduced network utilisation.
  • TODO(security): pre-shared list of trusted keys. Whether a packet should be accepted is determined by whether it was signed by a trusted node. In version 0.1 we define a list of trusted nodes at the outset, and nodes can not be subsequently added to the list of trusted nodes. Each trusted node has a pub/priv RSA 256 key pair.

interest packet: An interest packet is just a signed JWT. We do not define any JWT headers in version 0.1. The payload of the JWT contains the mandatory fields: "type", "data_name", "requestor_id". The optional fields in the payload are: "created_at", "expires_at"

Interest Request Format:

In version 0.1 of g17icn interest requests are send over HTTP POST requests. A peer can listen permanently for clients who want to propagate interest requests. TODO:

  • use hash of public key instead of full public key
HTTP 1.1
x-tcdicn-hop: 0
base64(<jwt-headers>
.{
    "type": "interest",
    "data_name": "/temperature/(129,30)" ,
    "requestor_public_key": "------PUBLIC KEY-----. .....",
    "created_at": 1231234.123,
}.
<signature>)
HTTP 1.1
x-tcdicn-hop: 0
base64(<jwt-headers>
.{
    "type": "satisfy",
    "data_name": "/temperature/(129,30)" ,
    "data": {"unit": "celsius", "value": 8},
    "sender_public_key": "------PUBLIC KEY-----. .....",
    "created_at": 1231234.123,
}.
<signature>)

Interest Table Format

{ "data_name":request_type/(location), "request_type":flag # when flag=1 represented this request from other nodes, 0 represented from itself. "waiting_list":[] #nodes index waiting for this request }

Neighbour

neighbour list should include the port number and its location.locals

Node

self.task_id = task_id # ? self.logger = logging.getLogger() self.jwt = g17jwt.JWT().init_jwt(key_size=32) self.PIT = {} self.FIB = {} self.emulation = emulation self.location="124.0.0" # represented the node location self.server = HTTPServer() self.neighbour=[] # represented the node neighbour

Cache

self.CIS = {} self.PIT = {} self.FIB = {}