1-6-1. Decentralized Network of FRUITS

The following diagram is often used as a commonly accepted image of a blockchain, but there are various physical problems to be solved in order to actually build a decentralized network.

For example, if we try to construct a decentralized network from the conceptual diagram shown in Figure 9(A), it will look like Figure 9(B) due to the infrastructure such as transactions between each node (terminal: node). In Figure 9(B), a small central node exists in the center of each group (node with more than one contact point), which is the starting point when connecting distant nodes when the adjacent nodes are physically configured. As a result, the transaction to this node increases and the node becomes overloaded.

Figure 9: Reticulation and star diagram

In addition, there are transaction congestion problems.

For example, suppose there is an active transaction between the area of node A (Figure 10: a0 to a4) and the area of node B (Figure 10: b0 to b2). In this case, if the mining process is concentrated, the network around the small central node (the starting point) will be overloaded, and if no measures to avoid this overload are taken, the transactions will increase and become congested.

Avoiding these node overloads and transaction congestion is the solution to speeding up distributed networks, and there are many ways to avoid this in Blockchain.

Figure 10: Avoiding transaction congestion by detour and time difference

We are solving this problem by using an AI (Artificial Intelligence) Genetic Algorithm (GA) and Neural Network (NN) (Shown in Figure 10(B)).

In the search for the shortest path, the line connecting each node is regarded as a road, and environmental factors and transaction types are quantified and patterned, and the information is accumulated as knowledge and shared with each node. In order to cope with the next possible load (traffic jam), the optimal pattern is always sent to each small central node (a0, b0), and the small central node sends the data to each node where the mining work is actively performed.

The small central node sends the data to each node where the mining work is actively performed, thereby improving the efficiency of the mining work.

Basically, the transaction between nodes is one-way as shown in Figure 11. However, as a whole, these transactions form a loop-like structure and are linked together to form a single Blockchain network. The network routes are assigned parameters as shown in Figure 11(B), and the route numbers of all networks are all unique numbers.

This unique number allows all nodes to share and access all data, even if each node does not have all the route information.

Figure 11: Parameter substitution for shortest path search.

As an even more efficient way to avoid traffic jams, we have adopted a unique method called radar scanning (Figure 12). We send a random token in between certain triggering signals, i.e., the mining process. Periodically, each node communicates with each other, and each node in the data keeps approximate location information (its position in the FRUITS Blockchain network), finds the shortest path to its neighborhood, and sorts the information. This sorting method incorporates random intersections as shown in Figure 12 to have knowledge of node areas and routes that change in real time.

Figure 12: Relationship between radar scanning and GA intersection settings

Figure 13: Information sharing and routing among Small Central Nodes

This allows each node to select the optimal transaction while avoiding traffic jams.