Scholarlink

<p>Healthcare IoT systems have brought in significant revolution in the healthcare sector, making it more easy and simple to offer services surpassing all the geographical barriers. But, these systems generate immensely huge amounts of data which also include redundant information like network log, biometric data, storage and transmission related metadata etc. This redundant information requires a suitable and time-based removal as it takes a lot of storage unnecessarily. This work proposes a combined method that joins the Controlled Cut-point Identification Algorithm with Reservoir Sampling - based Bloom’s Filter (RSBF) which offers an optimized data deduplication for resource - constrained environments like Healthcare IoT systems. The controlled cut-point identification helps in identifying context aware boundaries in the data stream, which helps in easy identification of redundant data blocks, The reservoir sampling based bloom filter is then applied across multiple tiers of the data - like device, fog and cloud, detecting and removing duplicate data blocks. The proposed CCIA - RSBF model is validated over two prominent healthcare IoT network datasets - WUSTL EHMS 2020 and CICIoMT2024. A comparative analysis of the performance with other state-of-the-art approaches like Asymmetric Extremum(AE), Rapid Asymmetric Maximum(RAM), Stable Bloom Filter(SBF), Parallel Dedup, Locality Sensitive Hashing Bloom Filter (LShBloom), Multi-Pattern Guarded Deduplication (MP-Guard) and Speed Dedup has been carried out which proved that the CCIA-RSBF model had better performance levels in terms of deduplication ratio, false positive ratios and energy consumption aspects. With the WUSTL EHMS 2020 dataset the results achieved were respectively 92.65% in deduplication ratio, 0.79 false positive rate, and an energy saving upto 65%. While working on CICIoMT2024 dataset, the results achieved were respectively 93.12% in deduplication ratio, 0.81 false positive rate, and an energy saving upto 60%. The proposed model has shown superior levels of performances with both datasets. The study also analyses the future enhancements for the proposed model, thereby suggesting future research directions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A hierarchical framework for healthcare IoT deduplication: context-aware chunking and dynamically-tuned multi-tiered filtering of network metadata

  • 期刊论文

  • 发表日期：

    2026年6月30日

查看全文
  • Arul Treesa Mathew,
  • Prasanna Mani

摘要

Healthcare IoT systems have brought in significant revolution in the healthcare sector, making it more easy and simple to offer services surpassing all the geographical barriers. But, these systems generate immensely huge amounts of data which also include redundant information like network log, biometric data, storage and transmission related metadata etc. This redundant information requires a suitable and time-based removal as it takes a lot of storage unnecessarily. This work proposes a combined method that joins the Controlled Cut-point Identification Algorithm with Reservoir Sampling - based Bloom’s Filter (RSBF) which offers an optimized data deduplication for resource - constrained environments like Healthcare IoT systems. The controlled cut-point identification helps in identifying context aware boundaries in the data stream, which helps in easy identification of redundant data blocks, The reservoir sampling based bloom filter is then applied across multiple tiers of the data - like device, fog and cloud, detecting and removing duplicate data blocks. The proposed CCIA - RSBF model is validated over two prominent healthcare IoT network datasets - WUSTL EHMS 2020 and CICIoMT2024. A comparative analysis of the performance with other state-of-the-art approaches like Asymmetric Extremum(AE), Rapid Asymmetric Maximum(RAM), Stable Bloom Filter(SBF), Parallel Dedup, Locality Sensitive Hashing Bloom Filter (LShBloom), Multi-Pattern Guarded Deduplication (MP-Guard) and Speed Dedup has been carried out which proved that the CCIA-RSBF model had better performance levels in terms of deduplication ratio, false positive ratios and energy consumption aspects. With the WUSTL EHMS 2020 dataset the results achieved were respectively 92.65% in deduplication ratio, 0.79 false positive rate, and an energy saving upto 65%. While working on CICIoMT2024 dataset, the results achieved were respectively 93.12% in deduplication ratio, 0.81 false positive rate, and an energy saving upto 60%. The proposed model has shown superior levels of performances with both datasets. The study also analyses the future enhancements for the proposed model, thereby suggesting future research directions.