EBPC
Real Time Cloud Computing Framework for Tool Condition Monitoring
Jiduo Zhang1*, Robert Heinemann1 and Otto Jan Bakker1
1 Department of Mechanical and Aerospace Engineering, The University of Manchester
Highlights of our framework:
Exponential Backoff Traffic Control
Progressive traffic control mechanism to adaptively adjust data transmission rates based on network conditions, ensuring efficient and reliable communication between edge devices and the cloud.
Data minimisation
Compression by data shape, dimension and bit depth reduces payload size for transmission while retaining essential information for accurate tool condition monitoring.[1]
Producer-Consumer Task Scheduling
Efficient task scheduling mechanism that manages resource allocation effectively, ensuring timely processing of data and responsiveness of the tool condition monitoring system for all clients.
Dynamic Task Prioritisation
Self-adaptive prioritisation mechanism that dynamically adjusts task priorities based on real-time system conditions and requirements.
Exponential backoff In Client — interactive demo
Exponential Backoff Demo
The demo shows how the retry interval dynamically adapts: each failure doubles the delay (capped) and adds a small jitter. Use the network success slider to change the simulated probability of success per attempt.
Decision log (latest)
How it works
This interactive demo runs a continuous sender that measures simulated network round-trip times (RTT) and adapts the send interval in real time. A 'Network success' slider controls the probability that a request succeeds; when a request fails the sender immediately doubles the send interval (exponential backoff). When a request succeeds the demo uses the observed RTT to adapt the interval: if the observed RTT is more than twice the current send interval the interval is doubled; if the observed RTT is less than half the current send interval the interval is halved. Small random jitter is added to adjustments to avoid synchronized retries.
Parameters: baseDelay = 100ms, adaptive rules (failure → ×2; success → double if RTT > 2×interval, halve if RTT < interval/2), jitter up to 10% of the interval, cap = 20000ms.
Bit-depth demo
Bit-depth demo
Interactive bit-depth quantization demo showing the effect of reducing per-channel bit depth on signal quality. Use the slider to change bits per channel and the noise level.
How it works
This interactive demo allows you to explore the effects of bit depth reduction on signal quality. The slider lets you adjust the bit depth from 1 to 8 bits per channel, allowing you to see how reducing the bit depth affects the signal's appearance. This is particularly useful for understanding how data compression can be achieved by reducing the number of bits used to represent each feature in the signal.
Note: Reducing bit depth can lead to a loss of precision information and may introduce visual artifacts, especially at very low bit depths. When the quantized noise exceeds the SNR of features in the original signal, features can become interfered, leading to a degradation in perceived quality.
Check the Minimum Sufficient Unit Theory for the compression from the aspect of data shape.
Producer / Consumer — interactive demo (comparison)
Compare the two demos side-by-side. The right-hand demo is configured with concurrency = 1 to simulate a traditional architecture that awaits the completion of each task before starting the next.
Concurrency = 4 (default)
This demo simulates a producer adding messages to a queue and consumers processing them. Toggle between a bounded queue (drops when full), a hard buffer (never drops) or a blocking producer to compare behaviour. A sparkline shows recent queue length.
Concurrency = 1 (Traditional architecture)
This demo simulates a producer adding messages to a queue and consumers processing them. Toggle between a bounded queue (drops when full), a hard buffer (never drops) or a blocking producer to compare behaviour. A sparkline shows recent queue length.
Deploy worldwide for Tool Condition Monitoring with minimal delay
Connectivity points (weighted)
Conclusions
- We proposed a novel cloud computing framework EBPC for real-time tool condition monitoring, which incorporates exponential backoff-based traffic control, data minimisation, producer-consumer task scheduling, and dynamic task prioritisation.
- Our framework effectively addresses challenges such as network congestion, data overload, and resource contention, ensuring efficient and reliable tool condition monitoring.
- Experimental results demonstrate that EBPC outperforms traditional architectures in terms of latency, throughput, and resource utilisation, making it a promising solution for real-time tool condition monitoring in manufacturing environments.
Jiduo would like to acknowledge the continuing contribution to this research by China scholarship council.
Statement for the unprofessional revision
I would like to appreciate the reviewers for their time and effort in reviewing my paper. However, it is extremely regrettable one of the reviewer’s recommendations for rejection based on points that do not originate from our manuscript.
If you have any questions or suggestions, please feel free to contact me. Except the following people
- Garima Nain
- Kiran Kumar Kumar Pattanaik
- Girish Kumar Sharma