In this paper, we address the performance analysis of Direct Acyclic Graph (DAG)-based cryptocurrency on the basis of its native structure. Several attempts on existing blockchain-based cryptocurrencies to resolve the limitation of the transaction processing rate has resulted in numerous form of solutions, such as side-chain, off-chain or data sharding. While those efforts are dedicated to overcome the limitation that the one-dimensional data structure has, a two-dimensional form of distributed ledger as known as DAG-based cryptocurrency has been emerged. Comparing with blockchain-based cryptocurrencies, DAG-based cryptocurrencies feature relatively faster issuance and confirmation of the transactions due to the individual processing method of the transactions. However, existing DAG-based cryptocurrencies also suffer from the instability of the performance, since the confirmation or validation delay can be varied by network condition such as transaction incoming rate. We derive the generic performance model of DAG-based cryptocurrency while varying the parameters. From the derived models, we propose some novel strategies to improve the performance of the DAG-based cryptocurrency system which encourages the participants of the cryptocurrency (both traders and nodes) to boost the transaction processing speed while keeping the ledger healthy. We show the effectiveness of our proposal by numerical statistics obtained by simulation, which results in maximum 46 percent of increase in our case.