Lossless Data Compression for Time-Series Sensor Data Based on Dynamic Bit Packing.

Area of Science:

• Data Compression
• Signal Processing
• Computer Science

Background:

• Traditional bit packing methods often lead to space wastage, especially with data containing outliers or multidimensional variations.
• Efficient compression is crucial for handling large datasets in AI training and data science predictive analysis.
• Optimizing storage and network bandwidth is essential for transmitting large volumes of sensor data.

Purpose of the Study:

• To introduce a novel lossless bit depth compression (BDC) technique.
• To maximize compression ratios by dynamically determining pack size based on sensor data patterns.
• To reduce spatial inefficiency caused by outliers in time-series data.

Main Methods:

• The proposed bit depth compression (BDC) technique dynamically determines pack size based on bit depth level patterns in sensor data.
• It performs bit packing, adapting to data characteristics like outliers and multidimensional variations.
• This lossless method is designed for applications in artificial intelligence and data science.

Main Results:

• The BDC method achieved an average compression ratio improvement of 30%, with a maximum of 247% compared to other algorithms.
• It demonstrated significant improvements in energy consumption for data transmission, with an 18% average reduction (up to 34%) via Bluetooth.
• The technique effectively addresses spatial inefficiency in time-series data compression.

Conclusions:

• The proposed BDC technique offers a highly efficient lossless compression solution for sensor data.
• It significantly enhances storage space savings and optimizes network bandwidth utilization.
• BDC is particularly beneficial for AI training data and data science predictive analysis due to its high compression ratio and energy efficiency.