5.4.1. Performance Improvements

Given the above numerical results in terms of processing time and running time resource usages, our PoENF consensus is more computationally efficient than the PoW-based methods. Such a lightweight property of PoENF is promising for reducing energy consumption in mining processes and can lower demands on system capability for participants. Thus, resource-limited IoVT devices can directly work as validators (miners) rather than depending on support from an intermediate consensus layer by outsourcing mining tasks on fog networks or cloud servers. Compared with these hardware dependent solutions, such as REM based on Intel SGX and PoR requiring large local storage, our PoENF consensus relies on a platform independent algorithm to extract ENF-containing multimedia signals from recordings as ENF proofs. Therefore, it is promising to address heterogeneity issues as we integrate blockchain technology with IoVT systems that include multiple non-standard platforms.

EconLedger achieves communication efficiency by executing consensus protocol within a random selected PoENF committee. Such a small scale consensus network imposes low levels of data transfer overhead on IoVT systems at the network of edge, which has limited bandwidth. In addition, communication complexity for each validator is linearly scaled to PoENF committee size, as shown in Figure 6. Thus, limited data transmission also means lower energy consumption on devices during communication handling tasks. Unlike non-scalable BFT-based solutions that rely on a pre-fixed set of validators, EconLedger aims to improve scalability by requiring a randomly elected consensus committee to delegate other nodes of the network. As a tradeoff, EconLedger is actually a partially decentralized blockchain network.

In EconLedger, raw data are saved into off-chain storage deployed on a DDB network, while only references of data are encapsulated into transactions that are finalized on distributed ledger (on-chain storage). As a reference is a fixed length of hash value disregarding format or size of the source data, such light transactions can be used to verified complicated data in use over IoVT systems, such as multimedia recordings, contextual information, and trained models, etc. Moreover, each tx has fixed and small size such that a block can record more txs. As a result, the txs rate increased given that the block confirmation time is stable.
