Apparatus and method for adaptively managing sharded blockchain network based on deep Q network

A method for managing sharded blockchain network, including the steps of:

• obtaining state information (St) including an inter-node data transmission rate (R) and computing resources (c) of each node in a current epoch (t) of a shard-based blockchain network in which N nodes are distributed to at least one shard, a consensus process record (H) of each node in each shard and a ratio of malicious nodes ();
• estimating at least one action (A) each including a block size (B), a block interval (TI) and the number of shards (K) corresponding to the state information (St) by using an artificial neural network in which a pattern estimation method has been previously learned;
• selecting one action (At) among the estimated at least one action (A) that maximizes a processing speed (transactions per second:
• hereinafter referred to as TPS) while satisfying a first constraint that the latency (Tlatency) is less than or equal to a successive block interval (uTI) and a second constraint that the number of shards (K) is within a pre-designated maximum number of secure shards () required to maintain security;
• and
• applying the block size (B), block interval (TI) and number of shards (K*) of the selected action (At) to the shard-based blockchain network.