Federated Learning Management & Management Dashboard (FLmon).
This dashboard was created to monitor and directly manage the performance of federated learning models.
Information required for the dashboard is stored and used in the sqlite3 database, indicating that it is still an early version.
The dashboard communicates with the user's federated learning server using the gRPC communication protocol.

1. Performance database creation. (same environment as user's FL server)
2. Insert performance data into database.
3. Run server/FLmonServer.py (same environment as user's FL server)
4. Run index.js (desired FLmon environment)
5. Start federated learning!

• FLmon is located in the different environment as user's federated learning server and is a tool for federated learning users.

• index.db (Save the unique name of the client participating in federated learning by matching the index.)

CREATE TABLE ClientID (
    id INTEGER AUTO_INCREMENT PRIMARY KEY,
    clientname TEXT NOT NULL UNIQUE);

• learning.db (Save all data for performance monitoring.)

CREATE TABLE LearningRound (
    round INTEGER NOT NULL,
    clientid INTEGER NOT NULL,
    datasize TEXT NOT NULL,
    classsize TEXT NOT NULL);
CREATE TABLE LearningTime (
    round INTEGER NOT NULL,
    clientid INTEGER NOT NULL,
    uploadstarttime REAL NOT NULL,
    uploadendtime REAL NOT NULL);
CREATE TABLE DistributionTime (
    round INTEGER NOT NULL,
    clientid INTEGER NOT NULL,
    distributiontime REAL NOT NULL);
CREATE TABLE AggregationTime (
    round INTEGER NOT NULL,
    aggregationtime REAL NOT NULL);
CREATE TABLE LearningTrain (
    round INTEGER NOT NULL,
    clientid INTEGER NOT NULL,
    accuracy REAL NOT NULL,
    loss REAL NOT NULL,
    tloss REAL NOT NULL,
    trainingtime REAL NOT NULL);
CREATE TABLE Predictions (
    round INTEGER NOT NULL,
    clientid INTEGER NOT NULL,
    prediction REAL NOT NULL);
CREATE TABLE SelectionClient (
    round INTEGER NOT NULL,
    remove_client TEXT NOT NULL);

• Description of the data to be stored in the database.
• FLmon server-side gRPC server transmits performance data to FLmon server-side gPRC server at the request of FLmon server.

• Manage performance data for the purpose of resolving 4 anomalies.
  • Overfitting Problem
  • Heterogeneous Data Problem
  • Long Training Time Problem
  • Anomaly Data Problem
• Select client from FLmon server and forward to FL server side gRPC server.

• Federated Learning (gRPC)
• Federated Learning (WebSocket)
• Federated Learning (DL4J - mobile)