Personalized privacy preserving edge caching framework for recommended content distribution

Abstract

The invention relates to a personalized privacy preserving edge caching framework for recommended content distribution. The invention provides a novel edge caching framework aiming at the problems that the privacy protection requirement of a user is neglected when personalized recommendation is provided by an existing edge caching mechanism, so that the privacy leakage risk is caused, and a fixed privacy protection level cannot adapt to personalized privacy preferences of the user. The framework allows the user to dynamically adjust the privacy protection level according to personal privacy preferences, and noise of a corresponding level is added to a user request record by using a differential privacy technology so as to protect the privacy of the user. Meanwhile, the invention further provides a cache decision algorithm based on the return flow, and the cache hit rate is optimized by analyzing a user request mode in real time, dynamically adjusting a cache strategy and optimizing the cache hit rate. The framework of the invention comprises a content database, a recommendation server, an edge cache server and a user, and the edge cache server executes personalized privacy calculation and cache decision according to privacy setting and return flow data of the user. By means of the method, the privacy of the user is protected, meanwhile, noise added into the user data is reduced, the quality of a recommendation system is improved, the caching efficiency is improved, and the overall performance of the system is enhanced. The method is suitable for a distributed system which needs to provide personalized content recommendation and caching while protecting the privacy of the user, and particularly has important significance for improving the user experience and the system performance under the background of rapid development of the mobile internet and the big data technology.

Description

Technical Field

[0001] The present invention relates to the fields of computer science, network technology and information security, and in particular to a personalized privacy protection edge cache framework for recommendation content distribution. The framework aims to improve the efficiency of content distribution while protecting the personalized privacy needs of users. Background Art

[0002] With the rapid development of mobile Internet and big data technology, the information age we live in is undergoing a profound transformation. In this transformation, the recommendation system, as an important tool for content distribution, has become one of the key technologies to improve user experience and system performance. By analyzing the user's historical behavior, the recommendation system can predict and provide personalized content recommendations, which greatly enriches the user experience and brings higher user stickiness and conversion rate to service providers.

[0003] In order to further improve the efficiency and response speed of content distribution, edge caching technology came into being. By storing content at the edge of the network closer to the user, edge caching effectively reduces the delay of data transmission, reduces the load on the central server, and improves the availability and reliability of services. In data-intensive applications such as video streaming, online games, and e-commerce, the application of edge caching technology has become a standard practice in the industry.

[0004] However, while personalized recommendation systems provide convenience, they also bring challenges to user privacy protection. In order to provide accurate recommendations, these systems need to collect and process a large amount of user data, including sensitive information such as browsing history, search history, and purchase behavior. If the collection and use of this data is not properly managed and protected, it is very easy to cause the risk of user privacy leakage, resulting in a decrease in user trust in the service, and may even face legal and compliance challenges.

[0005] The existing edge caching mechanism has obvious deficiencies in privacy protection. Most systems adopt a fixed level of privacy protection. This "one-size-fits-all" approach ignores the personalized needs of users for the degree of privacy protection. For users with high privacy protection requirements, excessive noise addition will reduce the quality of recommendations and affect the user experience; while for users with low privacy protection requirements, too high a level of privacy protection may lead to a decrease in the quality of recommended content and cache hit rate, affecting the overall performance of the system. In addition, the existing edge caching mechanism lacks the ability to dynamically adapt to user request patterns. In the face of rapid changes in user behavior and diverse content requirements, static caching strategies are often difficult to adjust in a timely manner, resulting in a low cache hit rate and inability to effectively utilize limited cache resources, thereby reducing system performance and efficiency.

[0006] In view of the above problems, the present invention proposes a personalized privacy-preserving edge cache framework for recommended content distribution. The framework allows users to select the privacy protection level according to their personal privacy preferences by introducing differential privacy technology, thereby protecting privacy while reducing the impact of noise on the recommendation system and cache hit rate. At the same time, the present invention also proposes a cache decision algorithm based on backhaul traffic feedback, which analyzes user request patterns in real time and dynamically adjusts the cache strategy to optimize the cache hit rate, thereby achieving dual optimization of user privacy protection and system performance. Summary of the invention

[0007] The purpose of the present invention is to propose a personalized privacy-protected edge cache framework to resolve the contradiction between privacy protection and recommendation quality and cache hit rate in the prior art. The present invention allows users to select the privacy protection level according to their personal privacy preferences by introducing differential privacy technology, thereby protecting privacy while reducing the impact of noise on the recommendation system and cache hit rate. In addition, the present invention also proposes a cache decision algorithm based on backhaul traffic, which dynamically adjusts the cache strategy by real-time analysis of user request patterns to optimize the cache hit rate.

[0008] The technical solution adopted by the present invention comprises the following steps:

[0009] 1. User privacy preference setting: The present invention allows users to set privacy protection levels according to their personal privacy preferences. These levels correspond to different differential privacy budgets, which affect the amount of noise added to user data;

[0010] 2. Differential privacy processing: Differential privacy technology is used to add a corresponding level of noise to the user request record according to the privacy protection level set by the user to protect the user's privacy;

[0011] 3. Content recommendation and caching: Use the request records with added noise to recommend content, generate recommended content, and cache part of the content on the edge server;

[0012] 4. Backhaul traffic analysis: Collect and analyze backhaul traffic data to identify user request patterns and provide a basis for caching decisions;

[0013] 5. Dynamic cache policy adjustment: Based on the results of the return traffic analysis, the cache content on the edge server is dynamically adjusted to optimize the cache hit rate.

[0014] The present invention has the following advantages:

[0015] 1. Enhanced personalized privacy protection: By allowing users to customize the privacy protection level, the present invention provides a more flexible privacy protection mechanism to better meet the privacy needs of different users.

[0016] 2. Optimized recommendation quality: By finely controlling the noise addition in differential privacy, the present invention minimizes the negative impact on the accuracy of the recommendation system while protecting user privacy.

[0017] 3. Improved cache efficiency: By utilizing a cache algorithm based on backhaul traffic perception, the present invention can more accurately predict user needs, improve cache hit rates, and reduce unnecessary data transmission.

[0018] 4. Dynamic adaptability: The framework of the present invention can respond to changes in user behavior and request patterns in real time, providing a more flexible and responsive content distribution service.

[0019] 5. Improved system performance: By optimizing the cache strategy and improving the cache hit rate, the present invention reduces the load on the central server and improves the overall performance and response speed of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 : Architecture diagram of personalized privacy-preserving edge cache system for recommended content distribution;

[0021] Figure 2 : Working steps diagram of the caching algorithm based on backhaul traffic awareness. DETAILED DESCRIPTION

[0022] The present invention and technical solution are further described in detail below in conjunction with the accompanying drawings.

[0023] The system architecture of the present invention is as follows Figure 1As shown in FIG, the network includes a content database, a recommendation server, an edge cache server, and a user. The content database is responsible for storing all content items and distributing content to users according to user requests. The recommendation server generates personalized recommendations based on user request records. The edge cache servers are distributed in the network and are responsible for caching decisions, providing user request records to the recommendation server, distributing recommendation lists to users, and performing personalized privacy calculations. The specific workflow is as follows: First, the user sets the privacy protection level according to his or her privacy preference and notifies the corresponding edge cache server. These settings are stored in a mapping table to hide the user's real identity with an anonymous identity. When a user requests content, if the content has been cached in the edge cache server, it is directly provided to the user from the edge cache server. Otherwise, the content is obtained from the content database through the backhaul link, generating backhaul traffic. Then, the edge cache server performs differential privacy processing on the real request record according to the user's privacy budget, adds noise, and then sends the processed request record to the recommendation server with an anonymous identity. The recommendation server generates a personalized recommendation list based on the received request record and sends it to the user. Finally, the edge cache server uses a backhaul traffic-aware caching algorithm to regularly update its cached content to adapt to changes in user requests.

[0024] Figure 2 The working steps of the cache algorithm based on backhaul traffic awareness proposed in the present invention are shown in Figure 1. Take the cache update of the edge cache server in the tth cycle as an example. In the first step, based on the user statistics of the t-1th cycle, the edge cache server calculates three popularity indicators: local popularity Global differential popularity And the popularity of return traffic Among them, local popularity is calculated based on the request records of users connected to the edge cache server; global differential popularity is calculated based on the request records of all users obtained by the recommendation server with differential noise added; and return traffic popularity is calculated based on the return traffic sent by the content database. All three popularity are calculated based on the formula Perform calculations, is the number of requests for content n in the t-1th cycle. The second step is to calculate the sum of local popularity and global differential popularity. Then the sum of the popularity of all t-1 cycles is composed into a sequence: The third step is to calculate the average popularity of the t-1 period based on the sequence. The fourth step is to use the Hidden Markov Model (HMM) to predict the tth period of the sequence. Finally, the popularity is adjusted based on the return traffic calculation All content is sorted according to the adjusted popularity and the top ranked content is cached. Among them, λ∈[0,1] is a weight control parameter, and the best parameter is selected based on experiments.

Claims

1. A personalized privacy-preserving edge cache framework for recommended content distribution, characterized in that: The method comprises the following steps: Step A: Receive user request and set the privacy protection level according to the user's personalized privacy preference; Step B: using differential privacy technology to add noise to the user request record according to the privacy protection level; Step C: Use the request record with added noise in the content recommendation system to generate recommended content; Step D: Cache the content in the edge server according to its popularity and feed back the caching decision to the recommendation system; Step E: Analyze user request patterns based on return traffic and dynamically adjust cache strategies to optimize cache hit rates.

2. The method according to claim 1, characterized in that In step A, users can independently select the privacy protection level, thereby adjusting the degree of noise addition according to their personal privacy needs.

3. The method according to claim 1, characterized in that In step B, the implementation of differential privacy techniques ensures aggregate query protection of user data while allowing only the necessary amount of noise to be added.

4. The method according to claim 1, characterized in that: In step C, the recommendation system uses the privacy-preserving request records to predict user preferences and generate personalized recommendations.

5. The method according to claim 1, characterized in that In step D, the edge server dynamically adjusts the cache content based on user requests and system feedback to adapt to changes in user preferences.

6. The method according to claim 1, characterized in that In step E, by analyzing the return traffic, the system is able to identify and cache content that is frequently requested but not cached, thereby reducing the return traffic and improving cache efficiency.

7. A system based on the method according to any one of claims 1 to 6, characterized in that: The system includes: a request receiving unit, for receiving a service request from a user; a privacy setting unit for setting a privacy protection level according to a user's choice; A differential privacy processing unit for adding noise to user request records; A recommendation unit, used to generate recommendation content; a cache unit for caching recommended content on the edge server; A feedback analysis unit to analyze the return traffic and adjust the caching strategy.

Images