With the recent increase in usage of USIM card, transportation card, credit card, and smart card, it has become a commonplace to adopt Java card that can support multiple applications. Java card has the advantage that it can be used for multiple purpo...
With the recent increase in usage of USIM card, transportation card, credit card, and smart card, it has become a commonplace to adopt Java card that can support multiple applications. Java card has the advantage that it can be used for multiple purposes, but the downside in comparison to existing native card is that Java card is slower.
Many optimization methods have been proposed in order to overcome the drawback, but due to the significantly slower nature of smart card’s ability to write to memory compared to read speed, even the most efficient method of optimizing Java Virtual Machine or execution Code would not contribute significantly to the improvement. The comparative test on the execution rate per each Java card commands subcategorized between memory and non-memory related execution time has revealed, much of those commands taking relatively long time to complete were connected to memory related executions. For this reason, it is expected that optimizing the method of memory write would significantly improve performance.
In addition, smart cards have the property of guaranteeing atomicity of the memory even in the event of unexpected power-off. A good example is the case of electronic money, where the transaction data in process must be protected. Data backup method provides such protection where large data is written to the NVRAM resulting in performance degradation.
This paper proposes a type of cache model that uses RAM Buffer to optimize memory write and additionally, offers solutions to expected problems and expectation in performance gains. Increase in performance is obvious when RAM is used to store data instead of NVRAM, however, it is impractical to use this method due to the issue of data loss.
This paper proposes a method of enhancing performance by using RAM Buffer and two backup buffers while guaranteeing atomicity and protecting the transaction data, and additionally, describes the performance improvement resulting from the proposed implementation.