School of Science and Technology 科技學院
Electronic and Computer Engineering 電子工程學系

Retrieving Cost Minimization for Wireless Edge Caching Networks

Student Lau Pok Him
Programme Bachelor of Science with Honours in Cyber and Computer Security
Supervisor Dr. Yaru Fu
Year of Completion 2022


Wireless content caching is a technology that prefetching popular contents and placing them at the wireless edge facilities, such as base-stations, cloud server etc. Given that the users requested contents are cached by the network, they don't have to access the remote server. From this point of view, proactive caching is able to reduce the transmission delay and improve users' experience. Since a large number of contents exist in prevalent systems and the cache capacity of Base Stations is limited, caching decision becomes a vital aspect to be optimized.

In this project, it will be to perform the content placements at base-stations to minimize all users' content retrieving costs via modeling the hierarchical content fetching procedures. It will also be doing the analysis the optimality and time complexity for the design cache decision algorithms.

Demonstration Video


Due to the often unforeseen network congestion and limited backhaul capacity, it is becoming increasingly impractical for mobile users to access the desired content from the remote cloud, posing a major challenge in meeting the stringent requirements of latency-sensitive applications.

Since a large number of contents exist in prevalent systems and the cache capacity of Base Station is limited, caching decision becomes a vital aspect to be optimized. In this project, I will be to perform the content placements at base-stations to minimize all users' content retrieving costs via modeling the hierarchical content fetching procedures. It will also be doing the analysis the optimality and time complexity for the design cache decision algorithms.

In order to perform the content placements at base stations to minimize all users' content to retrieving costs via modeling the hierarchical content fetching procedures, I am preparing to focus on the following object to reach the goal (i.e. minimize the cost for wireless edge caching networks):

  • Built up the generic content-oriented wireless caching networks;
  • Formula the cost of wireless edge caching networks;
  • Design the algorithms for solving the formulated optimization problems;
  • Perform simulations to verify the proposed solutions.

Methodologies and Technologies used

In order to achieve the cost minimization for wireless caching networks, I will describe the average system cost of a generic content-oriented wireless caching networks model and show its dependency on recommendation and caching strategies. After that, I will formulate the cost minimization problem oriented toward shared caching and recommendation decisions, considering the constraints of each content provider’s cache capacity budget, each individual user’s recommendation size, and the quality of the recommendations.


Figure 1. The generic content-oriented wireless caching networks design



In this work, my goal is to minimize the total cost of all subscribers, taking into account the limitations of each Content Provider's cache capacity, the size of the recommendation, and the quality of each user’s recommendation.

Recommendation and Caching Decision Algorithms Design

In the previous section, I had analysis that P(1) is a non-convex problem with two challenging parts to be solve, it is develop a time-efficient suboptimal algorithms for joint optimization in this section.

Noticed that P(1) was split into two challenging sub-problems:

  • The caching decision problem (i.e. c)
  • The recommendation decision problem (i.e. b)

To make the problem be more solvable, I am preparing to use two algorithms to devise address to the abovementioned challenging sub-problems. They are: [31]

  • OOPA – one-by-one padding algorithm
  • TSSA – two-side swapping algorithm

An iterative optimization paradigm is established as a result of the two-fold analysis to accomplish the recommendation decision (b) and caching decisions (c) together in an iterative way. The following subsequence section will include the detailed approaches for optimizing the two categories of variables mentioned above.

Result and Discussion

In this part, extensive and detailed numerical simulations are carried out in this section to demonstrate the efficacy of the algorithm, enabled feature information prediction paradigm and to test the performance of the developed joint recommendation and caching optimization algorithm.

I will use the following scenario in this wireless caching network:

  • 4 Content Providers (i.e. M = 4)
  • 40 Users (i.e. K = 40)
  • 3 Themes (i.e. J = 3)
  • 20 Content Items (i.e. I = 20)

Note that:

  • Caching cost between user k and the CP m in terms of requesting item i is
  • The cost of user k in terms of retrieving content i from the cloud server is stipulated to be 1, i.e.
  • The recommendation acceptance probability of user k (i.e. xk) is a random variable with a uniform distribution drawn from a specified interval. To be more explicit, the considered intervals in the simulations are (0,0.5) and (0.5,1), respectively.
  • I’ll given that each CP has the same cache capacity for the sake of simplicity, i.e. for .

Several baseline systems are considered for the goal of comprehensive performance comparison, and they are distinguished as follows:

  • Without a recommendation, the top preferred caching [32]: In this approach, each Content Provider m caches the top desired items based on all users’ aggregated inherent preferences in order to fill up its storage entity, and no recommendations are made.
  • With heterogeneous recommendations, the top desired caching is: The caching choice is identical to the first technique in this system, except that each user is advised with their top favored things individually.
  • Top recommended caching with a consistent recommendation: In this situation, the caching approach is the same as in the previous two methods, but each user is given the top favored items based on the aggregated preferences of all users. [33]

To represent this performance, I employ the system caching cost throughout iterations. Figure 3 shows that the produced algorithm converges swiftly for all values of c. After the simulation by using MATLAB tools, I found that when the number of c is increasing, the system cost of the interaction number will be decrease, that means the system cost is direct to the interaction number inversely.

Furthermore, by comparing the  in fig 2a and  in fig 2b, I can see that the system cost of  is less than  for all number of c (in this case is c = 3, 4 and 5).

From fig 2b, if c = 5, the starting system cost at interaction number 2 is below 0.5, and the interaction number at 12 is nearest to 0, so it can believe that if the interaction number be much higher, the system will be the lower.

Base on the finding that I have found in this sub-section and the simulation, it can conclude that a high recommendation acceptance ratio leads to a low system cost, indicating that recommendations are successful.


Based on the above simulation result, it can see that under the anticipated preference distribution, the system cost performance of our created algorithm is comparable to that of the ideal situation.

Furthermore, when c increases, the average system caching cost of all the schemes falls, illustrating the efficacy of edge caching. Furthermore, all of the suggestion-based schemes outperform the no-recommendation policy by a large margin, demonstrating the effectiveness of recommendation.

Furthermore, across all the techniques, the proposed algorithm achieves the lowest average system caching cost for any given c. Furthermore, as c grows, the performance difference between the suggested scheme and the simplified scheme becomes less pronounced.

When comparing to the simulation result I found in fig 2, the result of trends is quite similar with the result found on fig 3.

In fig 3(b), the average system cost of caching policies with recommendations to that of caching policies without recommendations, the average system cost of caching policies with recommendation lowers dramatically.


The results are also similar to fig 3. But I can found that when R increase, the average system cost will also increase. That means that if the recommendation size in much higher, then the cost will be higher. This is because the caching policy is fixed for a recommendation decision. Aside from that, the cache size is limited.


In this project, I am aimed to retrieving Cost Minimization for Wireless Edge Caching Networks. Based on these topic, the average caching cost minimization problem was investigated in a general network context by considering undiscovered users feature information.

In this work, the cache size of the content provider m, the recommendation size R, were taken into consider. To make the problem tractable, I have used the OOPA (one-by-one padding algorithm) and TSSA (two-side swapping algorithm) to help me solve the very difficult mathematics NP-hard equation. The resulting joint decision issue was explored using the obtained users’ preference information. To be more explicit, the NP-harness of finding the best recommendation choice and cache placement options that reduce the average system cost was rigorously demonstrated.

After complete the simulation of Proposed Recommendation Decision Algorithm's Convergence and Caching Capacity and Recommendation Acceptance, it can conclude that a high recommendation acceptance ratio can leads to a low system cost and the average system cost of the no-recommendation approach remains constant throughout a range of recommendation and values.

Jonathan Chiu
Marketing Director
3DP Technology Limited

Jonathan handles all external affairs include business development, patents write up and public relations. He is frequently interviewed by media and is considered a pioneer in 3D printing products.

Krutz Cheuk
Biomedical Engineer
Hong Kong Sanatorium & Hospital

After graduating from OUHK, Krutz obtained an M.Sc. in Engineering Management from CityU. He is now completing his second master degree, M.Sc. in Biomedical Engineering, at CUHK. Krutz has a wide range of working experience. He has been with Siemens, VTech, and PCCW.

Hugo Leung
Software and Hardware Engineer
Innovation Team Company Limited

Hugo Leung Wai-yin, who graduated from his four-year programme in 2015, won the Best Paper Award for his ‘intelligent pill-dispenser’ design at the Institute of Electrical and Electronics Engineering’s International Conference on Consumer Electronics – China 2015.

The pill-dispenser alerts patients via sound and LED flashes to pre-set dosage and time intervals. Unlike units currently on the market, Hugo’s design connects to any mobile phone globally. In explaining how it works, he said: ‘There are three layers in the portable pillbox. The lowest level is a controller with various devices which can be connected to mobile phones in remote locations. Patients are alerted by a sound alarm and flashes. Should they fail to follow their prescribed regime, data can be sent via SMS to relatives and friends for follow up.’ The pill-dispenser has four medicine slots, plus a back-up with a LED alert, topped by a 500ml water bottle. It took Hugo three months of research and coding to complete his design, but he feels it was worth all his time and effort.

Hugo’s public examination results were disappointing and he was at a loss about his future before enrolling at the OUHK, which he now realizes was a major turning point in his life. He is grateful for the OUHK’s learning environment, its industry links and the positive guidance and encouragement from his teachers. The University is now exploring the commercial potential of his design with a pharmaceutical company. He hopes that this will benefit the elderly and chronically ill, as well as the society at large.

Soon after completing his studies, Hugo joined an automation technology company as an assistant engineer. He is responsible for the design and development of automation devices. The target is to minimize human labor and increase the quality of products. He is developing products which are used in various sections, including healthcare, manufacturing and consumer electronics.

Course Code Title Credits
  COMP S321F Advanced Database and Data Warehousing 5
  COMP S333F Advanced Programming and AI Algorithms 5
  COMP S351F Software Project Management 5
  COMP S362F Concurrent and Network Programming 5
  COMP S363F Distributed Systems and Parallel Computing 5
  COMP S382F Data Mining and Analytics 5
  COMP S390F Creative Programming for Games 5
  COMP S492F Machine Learning 5
  ELEC S305F Computer Networking 5
  ELEC S348F IOT Security 5
  ELEC S371F Digital Forensics 5
  ELEC S431F Blockchain Technologies 5
  ELEC S425F Computer and Network Security 5
 Course CodeTitleCredits
 ELEC S201FBasic Electronics5
 IT S290FHuman Computer Interaction & User Experience Design5
 STAT S251FStatistical Data Analysis5
 Course CodeTitleCredits
 COMPS333FAdvanced Programming and AI Algorithms5
 COMPS362FConcurrent and Network Programming5
 COMPS363FDistributed Systems and Parallel Computing5
 COMPS380FWeb Applications: Design and Development5
 COMPS381FServer-side Technologies and Cloud Computing5
 COMPS382FData Mining and Analytics5
 COMPS390FCreative Programming for Games5
 COMPS413FApplication Design and Development for Mobile Devices5
 COMPS492FMachine Learning5
 ELECS305FComputer Networking5
 ELECS363FAdvanced Computer Design5
 ELECS425FComputer and Network Security5