Status Update Optimization for Information Freshness and Real-Time Tracking
Thesis event information
Date and time of the thesis defence
Place of the thesis defence
IT115
Topic of the dissertation
Status Update Optimization for Information Freshness and Real-Time Tracking
Doctoral candidate
Master of Science Saeid Sadeghi Vilni
Faculty and unit
University of Oulu Graduate School, Faculty of Information Technology and Electrical Engineering, CWC-Radio Technologies
Subject of study
Communication engineering
Opponent
Professor Carlo Fischione, KTH University
Custos
Associate Professor Hirley Alves, University of Oulu
Status Update Optimization for Information Freshness and Real-Time Tracking
This thesis explored different strategies to improve status update systems, which are essential for keeping information fresh in real-time communication systems. These systems are crucial in areas like the Internet of Things (IoT), wireless communications, and smart grids, where devices need to regularly update their status to ensure timely decision-making. The research mainly focused on making these systems more efficient, especially when faced with challenges such as unreliable communication channels and limited energy resources. Below is a breakdown of the main findings from each chapter, explaining the strategies and methods used to optimize these systems.
Chapter 2: HARQ-Based Multi-Source Status Update System
Chapter 2 of the thesis focused on a specific kind of status update system that uses HARQ (Hybrid Automatic Repeat reQuest) technology. HARQ is a method used to ensure data is transmitted correctly even when the communication channel is not perfect. This chapter dealt with a system where multiple sources send updates over this unreliable channel.
The goal was to reduce the number of transmissions needed while keeping the information fresh and up-to-date. The system was studied in two different environments: one where the system's conditions are predictable (known environment) and another where the conditions are uncertain (unknown environment).
In the known environment, where the system's behavior can be predicted, a specific transmission policy was designed using RVIA (Relative Value Iteration Algorithm) and a technique called bisection. Additionally, a simpler policy, LC-DT, was developed using the DPP method, which reduced the complexity of the system while maintaining its performance.
In the unknown environment, where the system’s conditions are uncertain, a learning-based approach was adopted. The system used DQL (Deep Q-Learning), an algorithm that allows the system to "learn" the optimal way to transmit updates by observing feedback from its environment.
The results showed that both the deterministic policies (for the known environment) and the learning-based policy performed very well. In fact, the policies improved performance by up to 40% compared to a baseline policy. This result demonstrated the potential of HARQ in enhancing the freshness of information in multi-source systems, even when the communication channel is unreliable.
Chapter 3: Time-Slotted Computation-Intensive Status Update System
Chapter 3 explored a system where the updates require significant computational resources. This chapter tested two fixed policies to determine when to take and transmit new samples, depending on the availability of servers. These policies help decide how the system manages incoming data and when to send it out.
The Zero-Wait-One Policy transmits data immediately when both servers are available.
The Zero-Wait-Blocking Policy blocks any incoming data from being transmitted if the server responsible for transmission is busy. In addition, an optimal control policy based on MDP (Markov Decision Process) was proposed. The goal of this policy was to minimize the age of the information (AoI) or ensure that the system is always transmitting fresh data. By comparing the fixed policies with the optimal one, it was found that the Zero-Wait-One Policy worked best when service rates were low, but as service rates increased, the Zero-Wait-Blocking Policy became more effective. In some cases, it even approached the performance of the optimal policy.
This chapter showed how to balance computational efficiency with the need for real-time updates, helping to determine the best strategy under different operating conditions.
Chapter 4: Energy Harvesting and Status Update System
In Chapter 4, the focus shifted to a system that harvests energy from its environment (such as from solar power or vibration) to transmit updates. The key challenge here was that both the forward transmission channel (sending data) and the feedback channel (receiving confirmation) could be unreliable. The system aimed to minimize distortion, which refers to the difference between the real state of the source and the transmitted information, while considering energy constraints.
To address this, the problem was modeled using a POMDP (Partially Observable Markov Decision Process), a method for situations where the system has limited knowledge about its current state. A transmission policy was proposed using the RVIA method. Additionally, a simpler, low-complexity policy was developed to make the system easier to implement, especially when computational resources are limited.
The results showed that the RVIA-based policy used a switching approach, meaning it decided when to transmit based on the channel conditions and how frequently the source state changed. When the channel was in good condition and the source changed often, the system transmitted less frequently, saving energy. The low-complexity policy performed similarly to the RVIA-based policy, providing a good balance between performance and computational efficiency.
In summary, this thesis proposed several innovative strategies to optimize status update systems, aiming to minimize the number of transmissions, improve the freshness of the transmitted information, and address challenges such as energy limitations and unreliable communication channels. The study combined both deterministic and learning-based policies, demonstrating how these approaches can significantly improve system performance, especially when using HARQ technology in multi-source systems.
The research also highlighted the importance of computational efficiency and energy harvesting in real-time systems, offering insights into how to balance the need for timely updates with available resources. By applying techniques like MDP, DQL, and RVIA, the thesis developed policies that can achieve near-optimal transmission strategies, which are critical for real-time applications such as IoT, wireless communications, and smart grids.
These findings contribute to the broader field of communication engineering by providing new methods for keeping information fresh and reliable in challenging conditions. The proposed solutions have significant implications for future systems, where data freshness and reliability are critical. Examples of such systems include autonomous vehicles, smart cities, and industrial automation, where decisions need to be made in real-time based on up-to-date information.
By combining theoretical models with practical implementation, this research offers valuable insights into designing more efficient, reliable, and real-time communication systems. These insights are crucial for building communication networks and devices that can effectively support the growing demands of modern technologies.
Chapter 2: HARQ-Based Multi-Source Status Update System
Chapter 2 of the thesis focused on a specific kind of status update system that uses HARQ (Hybrid Automatic Repeat reQuest) technology. HARQ is a method used to ensure data is transmitted correctly even when the communication channel is not perfect. This chapter dealt with a system where multiple sources send updates over this unreliable channel.
The goal was to reduce the number of transmissions needed while keeping the information fresh and up-to-date. The system was studied in two different environments: one where the system's conditions are predictable (known environment) and another where the conditions are uncertain (unknown environment).
In the known environment, where the system's behavior can be predicted, a specific transmission policy was designed using RVIA (Relative Value Iteration Algorithm) and a technique called bisection. Additionally, a simpler policy, LC-DT, was developed using the DPP method, which reduced the complexity of the system while maintaining its performance.
In the unknown environment, where the system’s conditions are uncertain, a learning-based approach was adopted. The system used DQL (Deep Q-Learning), an algorithm that allows the system to "learn" the optimal way to transmit updates by observing feedback from its environment.
The results showed that both the deterministic policies (for the known environment) and the learning-based policy performed very well. In fact, the policies improved performance by up to 40% compared to a baseline policy. This result demonstrated the potential of HARQ in enhancing the freshness of information in multi-source systems, even when the communication channel is unreliable.
Chapter 3: Time-Slotted Computation-Intensive Status Update System
Chapter 3 explored a system where the updates require significant computational resources. This chapter tested two fixed policies to determine when to take and transmit new samples, depending on the availability of servers. These policies help decide how the system manages incoming data and when to send it out.
The Zero-Wait-One Policy transmits data immediately when both servers are available.
The Zero-Wait-Blocking Policy blocks any incoming data from being transmitted if the server responsible for transmission is busy. In addition, an optimal control policy based on MDP (Markov Decision Process) was proposed. The goal of this policy was to minimize the age of the information (AoI) or ensure that the system is always transmitting fresh data. By comparing the fixed policies with the optimal one, it was found that the Zero-Wait-One Policy worked best when service rates were low, but as service rates increased, the Zero-Wait-Blocking Policy became more effective. In some cases, it even approached the performance of the optimal policy.
This chapter showed how to balance computational efficiency with the need for real-time updates, helping to determine the best strategy under different operating conditions.
Chapter 4: Energy Harvesting and Status Update System
In Chapter 4, the focus shifted to a system that harvests energy from its environment (such as from solar power or vibration) to transmit updates. The key challenge here was that both the forward transmission channel (sending data) and the feedback channel (receiving confirmation) could be unreliable. The system aimed to minimize distortion, which refers to the difference between the real state of the source and the transmitted information, while considering energy constraints.
To address this, the problem was modeled using a POMDP (Partially Observable Markov Decision Process), a method for situations where the system has limited knowledge about its current state. A transmission policy was proposed using the RVIA method. Additionally, a simpler, low-complexity policy was developed to make the system easier to implement, especially when computational resources are limited.
The results showed that the RVIA-based policy used a switching approach, meaning it decided when to transmit based on the channel conditions and how frequently the source state changed. When the channel was in good condition and the source changed often, the system transmitted less frequently, saving energy. The low-complexity policy performed similarly to the RVIA-based policy, providing a good balance between performance and computational efficiency.
In summary, this thesis proposed several innovative strategies to optimize status update systems, aiming to minimize the number of transmissions, improve the freshness of the transmitted information, and address challenges such as energy limitations and unreliable communication channels. The study combined both deterministic and learning-based policies, demonstrating how these approaches can significantly improve system performance, especially when using HARQ technology in multi-source systems.
The research also highlighted the importance of computational efficiency and energy harvesting in real-time systems, offering insights into how to balance the need for timely updates with available resources. By applying techniques like MDP, DQL, and RVIA, the thesis developed policies that can achieve near-optimal transmission strategies, which are critical for real-time applications such as IoT, wireless communications, and smart grids.
These findings contribute to the broader field of communication engineering by providing new methods for keeping information fresh and reliable in challenging conditions. The proposed solutions have significant implications for future systems, where data freshness and reliability are critical. Examples of such systems include autonomous vehicles, smart cities, and industrial automation, where decisions need to be made in real-time based on up-to-date information.
By combining theoretical models with practical implementation, this research offers valuable insights into designing more efficient, reliable, and real-time communication systems. These insights are crucial for building communication networks and devices that can effectively support the growing demands of modern technologies.
Last updated: 27.3.2025