Caching Problem Research Articles

On the Fundamental Limits of Device-to-Device Private Caching Under Uncoded Cache Placement and User Collusion

In the coded caching problem, as originally formulated by Maddah-Ali and Niesen, a server communicates via a noiseless shared broadcast link to multiple users that have local storage capability. In order for a user to decode its demanded file from the coded multicast transmission, the demands of all the users must be globally known, which may violate the privacy of the users. To overcome this privacy problem, Wan and Caire recently proposed several schemes that attain coded multicasting gain while simultaneously guarantee information theoretic privacy of the users’ demands. In Device-to-Device (D2D) networks, the demand privacy problem is further exacerbated by the fact that each user is also a transmitter, which appears to be needing the knowledge of the files demanded by the remaining users in order to form its coded multicast transmission. This paper shows how to solve this seemingly infeasible problem. The main contribution of this paper is the development of new achievable and converse bounds for D2D coded caching that are to within a constant factor of one another when privacy of the users’ demands must be guaranteed even in the presence of colluding users (i.e., when some users share cached contents and demanded file indices). First, a D2D private caching scheme is proposed, whose key feature is the addition of virtual users in the system in order to “hide” the demands of the real users. By comparing the achievable D2D private load with an existing converse bound for the shared-link model without demand privacy constraint, the proposed scheme is shown to be order optimal, except for the very low memory size regime with more files than users. Second, in order to shed light into the open parameter regime, a new achievable scheme and a new converse bound under the constraint of uncoded cache placement (i.e., when each user stores directly a subset of the bits of the library) are developed for the case of two users, and shown to be to within a constant factor of one another for all system parameters. Finally, the two-user converse bound is extended to any number of users by a cut-set type argument. With this new converse bound, the virtual users scheme is shown to be order optimal in all parameter regimes under the constraint of uncoded cache placement and user collusion.

Temporal-Spatial Collaborative Mobile Edge Caching With User Satisfaction Awareness

For mobile proactive edge caching problems, the hit action cannot always reflect the efficiency of the cache strategy unilaterally. Thus, the satisfaction information of users should be used to improve the intelligent cache strategy. Motivated by this, this paper focuses a caching scheme that jointly considers the cache evaluation both from the perspectives of the hit action and satisfaction of users (Algorithm 3). Accordingly, we first propose a requesting behavior prediction model (Algorithm 1) to handle the multiple content features. Then, a requesting behavior factor learning algorithm (Algorithm 2) based on the online gradient descent (OGD) with regulations is proposed to obtain the estimation of the parameters of the model in Algorithm 1. Further, the probabilistic tensor factorization learning (PTFL) model with collaborative filtering and Markov Chain Monte Carlo (MCMC) procedure are introduced for the learning process. Finally, the proposed the caching Algorithm 3 based on the hit action prediction and the collaborative temporal-spatial satisfaction learning scheme can be performed. Moreover, we theoretically combine the upper limit performance of user request prediction errors with the lower limit of user satisfaction of edge cache. The experimental results verify the effectiveness of the proposed algorithm.

Joint Transcoding- and Recommending-Based Video Caching at Network Edges

Edge caching can significantly improve the quality of wireless video services by deploying cache servers at network edges. Recently, video conversion and recommendation have been introduced to improve the caching performance at the edges. Specifically, they work to produce lower quality versions of videos via video converting (for the former) or provide alternative similar videos when requested videos are not available by using video recommendation (for the latter). However, existing work in this aspect has utilized these two techniques separately, which largely limit their capabilities in providing improved video services. In this article, we study how to jointly utilize these two techniques in edge caching for improved caching performance. The objective is to maximally reduce the video delivery delay while satisfying users’ requirements. We first formulate the optimal video caching problem in this case and derive its NP-hardness. We, then, propose an effective transcoding- and recommending-based caching algorithm (TRBA). The TRBA works in a greedy manner to iteratively buffer the most valuable video versions, one video version each time, until no more video version or cache space is available. We define the value of a video version as the reduced delivery delay if this version were buffered divided by the extra cache space required for its buffering. The computational complexity of TRBA is deduced as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$O(|V|^3|Q|^3)$</tex-math></inline-formula> , where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$|V|$</tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$|Q|$</tex-math></inline-formula> represent the total number of videos and the number of versions per video, respectively. Numerical results demonstrate that, compared with existing caching algorithms, TRBA can significantly improve the caching performance.

Cooperative Edge Caching Based on Temporal Convolutional Networks

With the rapid growth of networked multimedia services in the Internet, wireless network traffic has increased dramatically. However, the current mainstream content caching schemes do not take into account the cooperation of different edge servers, resulting in deteriorated system performance. In this paper, we propose a learning-based edge caching scheme to enable mutual cooperation among different edge servers with limited caching resources, thus effectively reducing the content delivery latency. Specifically, we formulate the cooperative content caching problem as an optimization problem, which is proven to be NP-hard. To solve this problem, we design a new learning-based cooperative caching strategy (LECS) that encompasses three key components. Firstly, a temporal convolutional network driven content popularity prediction model is developed to estimate the content popularity with high accuracy. Secondly, with the predicted content popularity, the concept of content caching value (CCV) is introduced to weigh the value of a content cached on a given edge server. Thirdly, an novel dynamic programming algorithm is developed to maximize the overall CCV. Extensive simulation results have demonstrated the superiority of our approach. Compared with the state-of-the-art caching schemes, LECS can improve the cache hit rate by 8.3%-10.1%, and reduce the average content delivery delay by 9.1%-15.1%.

Caching with Time Windows and Delays

We consider two generalizations of the classical weighted paging problem that incorporate the notion of delayed service of page requests. The first is the (weighted) paging with time windows (\sf PageTW) problem, which is like the classical weighted paging problem except that each page request only needs to be served before a given deadline. This problem arises in many practical applications of online caching, such as the “deadline” I/O scheduler in the Linux kernel and video-on-demand streaming. The second, and more general, problem is the (weighted) paging with delay (\sf PageD) problem, where the delay in serving a page request results in a penalty being added to the objective. This problem generalizes the caching problem to allow delayed service, a line of work that has recently gained traction in online algorithms (e.g., [Y. Emek, S. Kutten, and R. Wattenhofer, Proceedings of the 48th Annual ACM SIGACT Symposium on Theory of Computing, 2016, pp. 333--344; Y. Azar et al., Proceedings of the 49th Annual ACM SIGACT Symposium on Theory of Computing, 2017, pp. 551--563; Y. Azar and N. Touitou, Proceedings of the 60th IEEE Annual Symposium on Foundations of Computer Science, 2019, pp. 60--71]). We give $O(\log k\log n)$-competitive algorithms for both the \sf PageTW and \sf PageD problems on $n$ pages with a cache of size $k$. This significantly improves on the previous best bounds of $O(k)$ for both problems [Y. Azar et al., Proceedings of the 49th Annual ACM SIGACT Symposium on Theory of Computing, 2017, pp. 551--563]. We also consider the offline \sf PageTW and \sf PageD problems, for which we give $O(1)$-approximation algorithms and prove APX-hardness. These are the first results for the offline problems; even NP-hardness was not known before our work. At the heart of our algorithms is a novel “hitting-set” LP relaxation of the \sf PageTW problem that overcomes the $\Omega(k)$ integrality gap of the natural LP for the problem. To the best of our knowledge, this is the first example of an LP-based algorithm for an online problem with delays/deadlines.

Reinforcement learning for cost-effective IoT service caching at the edge

In the edge computing environment, Internet of Things (IoT) application service providers can rent resources from edge servers to cache their service items such as datasets and code libraries, and thus significantly reducing the service request latency and the core network traffic. Since IoT service providers need to pay for the rented edge computing resources, it is essential to find a dynamical service caching strategy to minimize the service cost while optimizing the performance objective such as service latency reduction. However, most of the existing studies either overlooked the problem of collaborative service caching or failed to consider the system's long-term service cost and latency. In this paper, to address such a problem, we coordinate multiple edge servers to cache service items and formulate the collaborative service caching problem using a multi-agent multi-armed bandit model. Furthermore, we propose a utility-aware collaborative service caching (UACSC) scheme based on a multi-agent reinforcement learning. The UACSC scheme can coordinate multiple edge servers to make a dynamic joint caching decision, aiming at maximizing the system's long-term utility. To evaluate the performance of our proposed scheme, we implement four representative baseline algorithms and compare them with six different performance metrics. In addition, a real-world case study is also presented to demonstrate the effectiveness of the UACSC scheme. Comprehensive experimental results show that the UACSC scheme can effectively coordinate multiple edge servers to cache service items, and achieve higher service latency reduction and lower service cost compared with other baseline algorithms.

Joint User Association and Edge Caching in Multi-Antenna Small-Cell Networks

Caching popular contents at edge networks (such as small-cell base stations) has been proposed to deal with the ever-growing mobile traffic. At the meantime, recommendation system is able to shape user demands for further prompting caching gain. In this paper, we study a multi-antenna multi-cell edge network employing transmit beamforming with caching-aware recommendation and user association. We first establish a framework for the joint problem of beamforming, user association, content caching and recommendation to minimize the content transmission delay of mobile users, by specifying a set of necessary conditions for all four component functions of the network. The resulting optimization problem corresponds to a non-convex, multi-timescale, and mixed-integer programming problem, which is hard to handle. To deal with the difficulty in solving the joint optimization problem by the direct formulation, we equivalently decompose it into three sub-problems. Then, we develop a computationally-efficient iterative algorithm to obtain the sub-optimal solution, where the three subproblems are tackled iteratively. Simulation results are conducted to demonstrate that the proposed algorithm can obtain lower transmission delay than baseline schemes.

Optimized Content Caching and User Association for Edge Computing in Densely Deployed Heterogeneous Networks

Deploying small cell base stations (SBS) under the coverage area of a macro base station (MBS), and caching popular contents at the SBSs in advance, are effective means to provide high-speed and low-latency services in next generation mobile communication networks. In this paper, we investigate the problem of content caching (CC) and user association (UA) for edge computing. A joint CC and UA optimization problem is formulated to minimize the content download latency. We prove that the joint CC and UA optimization problem is NP-hard. Then, we propose a CC and UA algorithm (JCC-UA) to reduce the content download latency. JCC-UA includes a smart content caching policy (SCCP) and dynamic user association (DUA). SCCP utilizes the exponential smoothing method to predict content popularity and cache contents according to prediction results. DUA includes a rapid association (RA) method and a delayed association (DA) method. Simulation results demonstrate that the proposed JCC-UA algorithm can effectively reduce the latency of user content downloading and improve the hit rates of contents cached at the BSs as compared to several baseline schemes.

Proactive Content Caching Based on Actor–Critic Reinforcement Learning for Mobile Edge Networks

Mobile edge caching/computing (MEC) has emerged as a promising approach for addressing the drastic increasing mobile data traffic by bringing high caching and computing capabilities to the edge of networks. Under MEC architecture, content providers (CPs) are allowed to lease some virtual machines (VMs) at MEC servers to proactively cache popular contents for improving users’ quality of experience. The scalable cache resource model rises the challenge for determining the ideal number of leased VMs for CPs to obtain the minimum expected downloading delay of users at the lowest caching cost. To address these challenges, in this paper, we propose an actor-critic (AC) reinforcement learning based proactive caching policy for mobile edge networks without the prior knowledge of users’ content demand. Specifically, we formulate the proactive caching problem under dynamical users’ content demand as a Markov decision process and propose a AC based caching algorithm to minimize the caching cost and the expected downloading delay. Particularly, to reduce the computational complexity, a branching neural network is employed to approximate the policy function in the actor part. Numerical results show that the proposed caching algorithm can significantly reduce the total cost and the average downloading delay when compared with other popular algorithms.

Multi-Antenna Coded Caching From a Placement Delivery Array for Shared Caches

The coded caching problem with shared caches where the server is equipped with multiple transmit antennas is considered. In a shared cache network, several users share a cache, but each user can access only a single cache. The fundamental limits of coded caching are known for the above setting under centralized uncoded placement. In the centralized case, to achieve the gains offered by coded caching, one requires a sub-packetization level which increases exponentially with the number of caches. The dedicated cache networks too had a similar issue, and placement delivery arrays (PDAs) were introduced as a solution to address the sub-packetization bottleneck of coded caching in the dedicated cache network. Our objective is to design coded caching schemes for shared caches with lower sub-packetization requirements than the existing schemes. By leveraging the PDA constructions, we propose a procedure to obtain new coded caching schemes for shared caches with reduced sub-packetization levels than the known schemes in both single and multi-antenna settings. The advantage of the proposed procedure is that all the existing PDA constructions can be transformed into coded caching schemes for shared caches, thereby reducing the sub-packetization level requirement. In addition, the proposed procedure also recovers the optimal caching schemes given by Parinello <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">et al.</i> by using a PDA corresponding to the Maddah-Ali Niesen scheme.

Cooperative multigroup multicast beamforming for cache‐enabled ultra‐dense low earth orbit satellite constellation networks

SummaryThe ultra‐dense low earth orbit (LEO) satellite constellation (UD‐LSC) networks are expected to be incorporated in future wireless networks, particularly 6G networks, to provide global wireless access with enhanced data rates. This paper investigates the downlink multigroup multicast transmission for UD‐LSC networks, in which LEO satellites provide the multicast service for ground users with full frequency reuse. Moreover, in‐network caching, as a key enabling technology of information‐centric networking, is brought into UD‐LSC networks to achieve both bandwidths saving and delay reduction. However, the multicast capacity is constrained by the signal to interference and noise ratio (SINR) of bottleneck users in each multicast group. Besides, the high‐speed movement of LEO satellites makes cache placement and updating more challenging than ground networks with immobile base stations. To track these problems, we first propose a cache‐enabled multi‐satellite cooperation framework and formulate the multicast transmission problem and the content caching problem. Then, to reduce the average content delivery delay (CDD) of ground users, we propose a cooperative multicast beamforming and transmission‐aware dynamic caching (CMB‐TDC) algorithm based on the sample average approach and successive convex approximation. Simulation results show that the proposed CMB‐TDC algorithm can improve the multicast capacity, and then the average CDD is reduced for UD‐LSC networks.

On the Optimal Memory-Load Tradeoff of Coded Caching for Location-Based Content

Caching at the wireless edge nodes is a promising way to boost the spatial and spectral efficiency, for the sake of alleviating networks from content-related traffic. Coded caching originally introduced by Maddah-Ali and Niesen significantly speeds up communication efficiency by transmitting multicast messages simultaneously useful to multiple users. Most prior works on coded caching are based on the assumption that each user may request all content in the library. However, in many applications the users are interested only in a limited set of content that depends on their location. For example, assisted self-driving vehicles may access super High-Definition maps of the area through which they are travelling. Motivated by these considerations, this paper formulates the coded caching problem for location-based content with edge cache nodes. The considered problem includes a content server with access to <inline-formula> <tex-math notation="LaTeX">${\mathsf N}$ </tex-math></inline-formula> location-based files (e.g., High-Definition maps), <inline-formula> <tex-math notation="LaTeX">${\mathsf K}$ </tex-math></inline-formula> edge cache nodes located at different regions, and <inline-formula> <tex-math notation="LaTeX">${\mathsf K}$ </tex-math></inline-formula> users (i.e., vehicles) each of which is in the serving region of one cache node and can retrieve the cached content of this cache node with negligible cost. Depending on the location, each user only requests a file from a location-dependent subset of the library. The objective is to minimize the worst-case load (i.e., the worst-case number of broadcasted bits from the content server among all possible demands). For this novel coded caching problem, we propose a highly non-trivial converse bound under uncoded cache placement (i.e., each cache node directly copies some library bits in its cache), which shows that a simple achievable scheme is optimal under uncoded cache placement. In addition, this achievable scheme is also proved to be generally order optimal within a factor of 3. Finally, we extend the coded caching problem for location-based content to the multiaccess coded caching topology originally proposed by Hachem <i>et al.</i>, where each user is connected to <inline-formula> <tex-math notation="LaTeX">${\mathsf L}$ </tex-math></inline-formula> nearest cache nodes. When <inline-formula> <tex-math notation="LaTeX">${\mathsf L}\geq 2$ </tex-math></inline-formula>, we characterize the exact optimality on the worst-case load.

A Novel Distributed Media Caching Technique for Seamless Video Streaming in Multi-Access Edge Computing Networks

Online video is anticipated to be the largest fraction of all mobile network traffic aside from the huge processing tasks imposed on networks by the billions of IoT devices, causing unprecedented challenges to the current network architecture. Edge caching has been proposed as a highly promising technology to overcome this challenge by placing computational and data storage resources at the network edge to reduce latency and backhaul traffic. However, the edge resources are heavily constrained in their storage and computational capacities as large-scale deployments mean fairly distributing resources across the network. Addressing this limitation, we propose an edge video caching scheme that dynamically caches the first part of popularity-ranked video files on Multi-Edge Computing Access Node (MAN) servers envisioned to achieve higher cache hit ratios, lower latencies, and lower backhaul traffic. The concept of Regionally Organized Clouds (ROCs) with sufficient resources for file caching and compute-intensive tasks was introduced, and a formulation of the edge caching problem as an Integer Linear Programming (ILP) problem was made. Additionally, this study proposes a file view-time threshold for each cached video aimed at reducing the resource wastage caused when buffered contents are abandoned. Comparative evaluations of the proposed show its excellent performance over FIFO, Greedy, LFRU and TLRU schemes.

Cooperative Activation and Caching Strategy for Low-Latency and Energy-Efficient Small-Cell Networks

Traffic load can be largely alleviated by content caching in densely deployed small-cell base stations (SBSs), which however require significant energy expenses for operation. In this letter, we investigate the cooperative activation and caching problem for dense small-cell networks with unknown user demands. A key issue is to make activation and caching decisions, which is proved to be NP-hard even if user demands are perfectly known. While existing works on joint optimization of activation and caching make decisions based on estimated user demands, we consider to learn activation and caching strategy in an online manner. We take into account the cooperation among SBSs. In particular, an active SBS that serves a user should share the observed rewards to the SBSs that are nearer to the user. In this way, an inactive SBS can keep learning the information of its surrounding users, and will become active again when its surrounding users have sufficient requests. The simulation results demonstrate the superior performance of the proposed strategy in terms of delay and energy consumption.

Delay-Optimal Edge Caching With Imperfect Content Fetching via Stochastic Learning

Caching popular contents in close proximity to the users can effectively reduce the latency in communication systems. Considering the imperfect content fetching from the original server, the caching management at the edge node should be determined according to not only the content popularity but also how difficult to fetch these content objects from their corresponding original servers. In this paper, we study the edge caching for minimizing the long-term average delay, where the fetching delay of the uncached contents is introduced. We construct a decision-theoretic framework for this delay-optimal content-caching problem, where the main obstacle is that the consideration of the imperfect content fetching breaks the Markovian property in the decision-theoretic framework. To overcome this obstacle, by analyzing the queue dynamics within the content fetching delay after the content object is removed from the cache, we transform the problem for meeting the Markovian property, and model it as an infinite horizon semi-Markov decision process (SMDP). Achieving the delay optimality of the caching problem needs to solve the Bellman equation of the SMDP, but it leads to the curse of dimensionality. We decompose the global optimality equation into several per-content optimality equations, and propose a low-complexity delay-optimal content caching algorithm by stochastic learning for each content. Finally, the simulation results show that our proposed algorithm achieves significantly lower delay than conventional caching algorithms.

Energy-Aware Collaborative Service Caching in a 5G-Enabled MEC With Uncertain Payoffs

Mobile edge computing (MEC) is an enabling technology for low-latency AI applications, by caching AI services originally deployed in remote data centers to 5G base stations in network edge. Due to limited computing resource of 5G base stations, not all services can be cached in base stations to meet the resource demands of user requests. Also, if the workload of a 5G base station reaches to its resource capacity, the energy consumption of the base station will be pushed up exponentially. To reduce the energy consumption and overcome resource limitations on base stations, an alternative is to allow the base stations to collaborate with each other to admit user requests. In this paper, we investigate the problem of collaborative service caching and request offloading between a 5G-enabled MEC and remote data centers, while meeting the quality of service (QoS) requirements of users, and resource capacities on base stations that are operated by multiple selfish network service providers. We aim to maximize the total payoff of all base stations. To this end, we first propose a two-stage optimization framework: In the first stage, we develop a mechanism that adopts a best-reply rule for dynamically distributed coalition formation. In the second stage, we propose a near-optimal payoff allocation method by devising a randomized algorithm with a provable approximation ratio. We then evaluate the performance of the proposed optimization framework by extensive experimental simulations. Simulation results show that the proposed framework outperforms its counterparts by achieving at least 30% higher payoff and 20% lower energy consumption of base stations.

Two-Layer Distributed Content Caching for Infotainment Applications in VANETs

For vehicular <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ad hoc</i> networks (VANETs), edge caching has attracted considerable research attention to maximize the efficiency and reliability of infotainment applications. In this article, we propose a two-layer distributed content caching scheme for VANETs by jointly exploiting the cache at both vehicles and roadside units (RSUs). Specifically, we formulate the content caching problem to minimize the overall transmission delay and cost as a nonlinear integer programming (NLIP) problem and propose an alternate dynamic programming search (ADPS)-based algorithm to solve it. In ADPS, we divide the original problem into three subproblems and then we use the dynamic programming (DP) method to solve each subproblem separately. To reduce the complexity, we further propose a cooperation-based greedy (CBG) algorithm to solve the large-scale original problem. Both numerical simulation results and experiments in the testbed show that the proposed caching scheme outperforms existed caching schemes, and the transmission delay and cost can be reduced by 10% and 24%, respectively, while the hit ratio can be increased by 30% in a practical environment, as compared to the popularity-based caching scheme.

Content Caching-Enhanced Computation Offloading in Mobile Edge Service Networks

Cache enhanced computation offloading as a novel offloading paradigm in mobile edge computing (MEC) can reduce more task execution latency than traditional computation offloading by reusing of computation offloading data. However, existing works only focus on the enhancement between computation offloading and data caching but ignore the competition for cache resources between them. To this end, in this paper, we propose a caching enhanced computation offloading algorithm in mobile edge service networks (MESN), by considering the cache resources competition. We formulate a joint optimization problem of content caching and cache-enhanced computation offloading. Furthermore, we give the optimal caching strategy to achieve the equilibrium between the resources competition. By our offloading algorithm caching strategy, the average response time of computation and content request tasks can get further reduction. In addition, we design two low time complexity algorithms, i.e., mixed caching algorithm and enhanced offloading algorithm, to solve the sub-problems, i.e., smart base station (SBS) caching sub-problem and computation offloading sub-problem, transformed by the original optimization problem. The simulation results show that our algorithms can quickly converge and our scheme can reduce 20.52&#x0025; average response time of all tasks at most compared with other schemes.

Data Caching Optimization With Fairness in Mobile Edge Computing

Mobile edge computing (MEC) provides a new computing paradigm that can overcome the inability of the traditional cloud computing paradigm to ensure low service latency by pushing computing power and resources to the network edge. Many studies have attempted to formulate edge data caching strategies for app vendors to optimize caching performance by caching the right data on the right edge servers. However, existing edge data caching approaches have unfortunately ignored fairness, which is an important issue from the app vendor&#x0027;s perspective. In general, an app vendor needs to cache data on edge servers to serve its users with insignificant latency differences at a minimum caching cost. In this paper, we make the first attempt to tackle the fair edge data caching (FEDC) problem. Specifically, we formulate the FEDC problem as a constraint optimization problem (COP) and prove its <inline-formula><tex-math notation="LaTeX">$\mathcal {NP}$</tex-math></inline-formula>-hardness. An optimal approach named FEDC-OPT is proposed to find optimal solutions to small-scale FEDC problems with integer programming technique. In addition, an approximate algorithm named FEDC-APX is proposed to find approximate solutions in large-scale FEDC problems. The performance of the proposed approaches is analyzed theoretically, and evaluated experimentally on a widely-used real-world data set against four representative approaches. The experimental results show that the proposed approaches can solve the FEDC problem efficiently and effectively.

Bandit Learning-Based Edge Caching for 360-Degree Video Streaming With Switching Cost

Virtual Reality (VR) and Augmented Reality (AR) applications are expected to be a key driver for the 5G network. The need of the MEC support for caching and transcoding of omnidirectional content has been identified as a major topic of research. In this paper, we study the optimal caching problem in an MEC-enabled VR system for tile-based viewport-adaptive 360-degree video streaming. For content caching, the replacement of the cached content inevitably incurs content switching cost, which is largely ignored in caching policy design. In this work, we aim to find the optimal caching policy to optimize the long-term transmission quality by considering both transmission latency and content switching cost. We apply the combinatorial multi-armed bandit (CMAB) theory to solve the problem with no a-priori knowledge on content popularity. Moreover, the CUCB with switching cost (CUCBSC) algorithm is adopted in this scenario. A transformation mechanism is designed to transform the generated single period optimization (SPO) problem into a multiple choice knapsack problem (MCKP). Rigorous theoretical analysis on the performance of the proposed algorithm is also provided. Finally, the effectiveness of the proposed learning based caching policy is confirmed by simulation results in terms of learning, hit-ratio, transmission and content switching delay.respondence) papers with a suitable choice of class options.