Optimal Caching Research Articles

Learning-Based Hierarchical Edge Caching for Cloud-Aided Heterogeneous Networks

Edge caching has emerged as a promising technique against latency caused by explosive growth of mobile data traffic through caching popular contents at the edge networks. However, the dynamically changing content popularity nature and limited caching capacity make it challenging to design an effective caching scheme to reduce latency. To solve this, a learning-based hierarchical edge caching (LHEC) scheme is proposed in this work. We first propose a novel deep learning architecture, namely Stacked Autoencoder-Long Short Term Memory Network (SAE-LSTMNet) to capture both the correlation of the request patterns among different content and the periodicity in time domain to improve the prediction accuracy of the content popularity. Then, to predict the popularity of these newly-added contents, a dynamic content catalog is introduced and a similarity-based content popularity prediction (SCPP) approach is proposed. Based on the content popularity prediction, a hierarchical edge caching optimization problem is formulated to minimize the average content downloading latency. Since the formulated problem is NP-hard and difficult to be solved, a low-complexity algorithm is proposed to obtain the near-optimal solutions. Simulation results show that the proposed content popularity prediction approach outperforms up to 6.36% in terms of the mean absolute error compared with the state-of-the-art methods and the proposed LHEC scheme reduces the average downloading latency at about 5.3%~7.9% compared with those existing caching schemes.

Optimal Delay-Outage Analysis for Noise-Limited Wireless Networks With Caching, Computing, and Communications

Performance assessment and optimization for networks jointly performing caching, computing, and communication (3C) has recently drawn significant attention because many emerging applications require 3C functionality. However, studies in the literature mostly focus on the particular algorithms and setups of such networks, while their theoretical understanding and characterization has been less explored. To fill this gap, this paper conducts the asymptotic (scaling-law) analysis for the delay-outage tradeoff of noise-limited wireless edge networks with joint 3C. In particular, assuming the user requests for different tasks following a Zipf distribution, we derive the analytical expression for the optimal caching policy. Based on this, we next derive the closed-form expression for the optimum outage probability as a function of delay and other network parameters for the case that the Zipf parameter is smaller than 1. Then, for the case that the Zipf parameter is larger than 1, we derive the closed-form expressions for upper and lower bounds of the optimum outage probability. We provide insights and interpretations based on the derived expressions. Computer simulations validate our analytical results and insights.

Lower energy consumption in cache-aided cell-free massive MIMO systems

Cell-free (CF) massive MIMO systems provide uniform and fast service to users by the cooperation of multiple distributed access points (APs) at the expense of high energy consumption (EC) due to the heavy use of fronthaul and backhaul links. To tackle this issue, we propose a cache-aided cell-free (CCF) system where all APs are deployed with edge cache to reduce the load on links. In this paper, a scalable total EC model of CCF systems is established with two parts: EC of APs (ECAP) and EC for cache transmission (ECCT). An improved partial zero-forcing precoding scheme is proposed, which can provide high spectral efficiency and transmit few control data to reduce ECAP greatly. Then a minimum ECCT strategy is formulated. We transform ECCT minimization problem into a series of geometric programming problems by the successive convex approximation and an ingenious equivalent transformation. The obtained optimal cache placement makes the ECCT of CCF systems lower than that obtained by maximum hit ratio, equal probability and most popular caching strategies. Simulation results show that the proposed CCF systems are much more energy efficient than CF systems.

Finding optimal non-datapath caching strategies via network flow

Flash and non-volatile memory (NVM) devices have only a limited number of write-erase cycles. Consequently, when employed as caches, cache management policies may choose not to cache certain requested items in order to extend device lifespan. In this work, we propose a simple single-parameter utility function to model the trade-off between maximizing hit-rate and minimizing write-erase cycles for such caches, and study the problem of developing an off-line strategy for deciding whether to write a new item to cache, and if so which item already in the cache to replace. Our main result is, OPTm, an efficient, network flow based algorithm which finds optimal cache management policy under this new setting.

Utility Maximization for IRS Assisted Wireless Powered Mobile Edge Computing and Caching (WP-MECC) Networks

This paper exploits an intelligent reflecting surface (IRS) assisted wireless powered mobile edge computing and caching (WP-MECC) network. In particular, an IRS is utilized to reflect energy signals from a power station (PS) to various IoT devices for energy harvesting during uplink wireless energy transfer (WET). These devices collect energy to support their own partially local computing for computational tasks and their offloading capabilities to an access point (AP), with the help of IRS via time or frequency division multiple access (TDMA or FDMA). The AP is equipped with a local cache connected with a MEC server via a backhaul link, which prefetches the data to facilitate edge computing capabilities. The maximization of a utility function is formulated to evaluate the overall network performance, which is defined as the difference between the sum of computational bits (offloading bits and local computing bits) and total backhaul cost. Due to multiple coupled variables, we first design the optimal caching strategy. Then, an auxiliary vector is introduced to coordinate the energy consumption of local computing and offloading, where its optimal solution can be achieved by an exhaustive search. Moreover, we utilize the Lagrange dual method and the Karush-Kuhn-Tucker (KKT) conditions to derive the optimal time scheduling for the TDMA scheme or the optimal bandwidth allocation for the FDMA counterpart in closed form. The IRS phase shifts are iteratively designed by employing the quadratic transformation (QT) and the Riemannian Manifold Optimization (RMO). Finally, simulation results are demonstrated to validate the network utility performance and confirm the advantage of the employment of IRS, the optimal IRS phase shift design and caching strategy, in comparison to the benchmark schemes.

Incentivizing Secure Edge Caching for Scalable Coded Videos in Heterogeneous Networks

Edge caching has been envisioned as a promising technology in heterogeneous networks (HetNets) to proximally cache (video) contents for mobile users. However, since caching contents inevitably consumes resources (e.g., power, bandwidth, storage), edge caching devices (ECDs) are unwilling to provide caching services. In addition, as the ECDs are usually deployed by untrusted third parties, the cached contents may be illegally accessed, which results in the mobile users’ privacy leakage. To efficiently address these problems, in this paper, we propose a novel secure edge caching scheme for video contents in HetNets. Specifically, to motivate the participation of ECDs, the Nash bargaining game is exploited to model the negotiations between the content provider and ECDs, where the optimal requested caching space of the content provider and the optimal caching price of each ECD are jointly analyzed. Apart from this, to protect the content secrecy, scalable video coding is employed to facilitate secure edge caching, where the ECDs are only utilized to cache the enhancement layers that cannot be independently decoded to reconstruct the original contents. Then, we formulate a non-convex 0-1 integer programming problem to optimize the enhancement layer caching on ECDs, and the distributed alternating direction method of multipliers (ADMM) is used to solve the problem optimally. Finally, simulation results show that the proposed scheme provides secure and efficient content caching for mobile users.

Joint Optimization of Transmit Beamforming and Base Station Cache Allocation in Multi-Cell C-RAN

This paper studies a downlink cloud radio access network (C-RAN) consisting of a centralized processor (CP) and multiple cells each one of which has one cache-enabled base station (BS) connected to the CP through wireless backhaul links. In such a network, we aim to determine sizes of the cache allocated to the BSs such that the long-term delivery time of the available files is minimized. Considering the two-stage nature of this problem, we formulate a beamforming optimization problem and a cache allocation problem. The purpose of the first one is to deliver un-cached portions of files to the BSs in the shortest possible time during the beamforming stage. The second problem aims to minimize long-term expectation of delivery time through optimal cache allocation during the cache allocation stage. We rigorously prove that the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">time-varying beamformers</i> can be assumed to be piece-wise constant functions of time. Based on this finding, the beamforming optimization problem does not appear to be amenable to a computationally efficient solution. Hence, we resort to the zero-forcing (ZF) beamforming approach to tackle the beamforming optimization problem. Using the results of the beamforming optimization problem, we prove that the cache allocation problem is a convex optimization problem; hence it can be solved by any convex optimization solvers. Simulation results show that the loss of optimality due to adopting ZF beamforming is negligible. Moreover, the superiority of our proposed cache allocation scheme over several other heuristic schemes including proportional cache size allocation is shown.

Energy-Aware Coded Transmission Strategy for Hierarchical Cooperative Caching Networks

With the rapid development of millions more edge base stations (EBSs) and billions of user devices in the future networks, the desire for low-energy system design and transmission strategy will be even more compelling. In this letter, we investigate the energy consumption issue of hierarchical cooperative caching networks (HCCNs) in which both EBSs and users are capable of storing content data in their local caches. A three-stage coded transmission strategy in HCCNs is proposed, and the closed expression of energy consumption of each stage is derived. In addition, we formulate a total energy consumption minimization problem with the constraint of EBSs’ cache sizes, and provide an algorithm to obtain the optimal cache placement matrix. We demonstrate by numerical results that, our optimized three-stage coded transmission strategy can achieve lower energy consumption than various multicast transmission strategies under the different users’ cache sizes.

A Time Pattern-Based Intelligent Cache Optimization Policy on Korea Advanced Research Network

Data is growing quickly due to a significant increase in social media applications. Today, billions of people use an enormous amount of data to access the Internet. The backbone network experiences a substantial load as a result of an increase in users. Users in the same region or company frequently ask for similar material, especially on social media platforms. The subsequent request for the same content can be satisfied from the edge if stored in proximity to the user. Applications that require relatively low latency can use Content Delivery Network (CDN) technology to meet their requirements. An edge and the data center constitute the CDN architecture. To fulfill requests from the edge and minimize the impact on the network, the requested content can be buffered closer to the user device. Which content should be kept on the edge is the primary concern. The cache policy has been optimized using various conventional and unconventional methods, but they have yet to include the timestamp beside a video request. The 24-h content request pattern was obtained from publicly available datasets. The popularity of a video is influenced by the time of day, as shown by a time-based video profile. We present a cache optimization method based on a time-based pattern of requests. The problem is described as a cache hit ratio maximization problem emphasizing a relevance score and machine learning model accuracy. A model predicts the video to be cached in the next time stamp, and the relevance score identifies the video to be removed from the cache. Afterwards, we gather the logs and generate the content requests using an extracted video request pattern. These logs are pre-processed to create a dataset divided into three-time slots per day. A Long short-term memory (LSTM) model is trained on this dataset to forecast the video at the next time interval. The proposed optimized caching policy is evaluated on our CDN architecture deployed on the Korean Advanced Research Network (KOREN) infrastructure. Our findings demonstrate how adding time-based request patterns impacts the system by increasing the cache hit rate. To show the effectiveness of the proposed model, we compare the results with state-of-the-art techniques.

Efficient implementation of singular value decomposition on shared memory computing systems

The problem of increasing the speed of big data analytics applications remains relevant. The solution may be to create new algorithms, or to increase the efficiency of existing implementations. Many studies are being conducted to optimize the operation of well-known algorithms in application to the computing complex as a whole, including the communications subsystem. At the same time, there is a need for efficient operation of the power of each node of the computing complex. In particular, the question of cache optimization and vectorization of a number of linear algebra algorithms that are part of the big data analytics software on parallel computing systems remains open. The purpose of this work is to increase the efficiency of the software implementation of a parallel linear algebra algorithm using the proposed optimization method. The proposed method was applied to the Jacobi method of singular decomposition of dense matrices. Computational experiments were carried out to determine the speedup of the proposed modification in relation to other software implementations. The results of computational experiments have shown the effectiveness of the proposed method

Joint optimization of task caching, computation offloading and resource allocation for mobile edge computing

Applications with sensitive delay and sizeable data volumes, such as interactive gaming and augmented reality, have become popular in recent years. These applications pose a huge challenge for mobile users with limited resources. Computation offloading is a mainstream technique to reduce execution delay and save energy for mobile users. However, computation offloading requires communication between mobile users and mobile edge computing (MEC) servers. Such a mechanism would difficultly meet users' demand in some data-hungry and computation-intensive applications because the energy consumption and delay caused by transmissions are considerable expenses for users. Caching task data can effectively reduce the data transmissions when users offload their tasks to the MEC server. The limited caching space at the MEC server calls for judiciously decide which tasks should be cached. Motivated by this, we consider the joint optimization of computation offloading and task caching in a cellular network. In particular, it allows users to proactively cache or offload their tasks at the MEC server. The objective of this paper is to minimize the system cost, which is defined as the weighted sum of task execution delay and energy consumption for all users. Aiming at establishing optimal performance bound for the system design, we formulate an optimization problem by jointly optimizing the task caching, computation offloading, and resource allocation. The problem is a challenging mixed-integer non-linear programming problem and is NP-hard in general. To solve it efficiently, by using convex optimization, Karmarkar's algorithm and the proposed fast search algorithm, we obtain an optimal solution of the formulated problem with manageable computational complexity. Extensive simulation results show that in comparison to some representative benchmark methods, the proposed solution can effectively reduce the system cost.

Caching in Narrow-Band Burst-Error Channels via Meta Self-Supervision Learning

We consider a coordinated small-cell network in narrow-band deep fading channels, where the transmissions from small base-stations (SBSs) to users usually encounter burst errors. Each SBS has limited cache-capacity via which partial contents can be cached to avoid being fetched from remote cloud and hence reduce the transmission delay. However, due to deep fading, the content transmission may occur numerous burst errors. To solve this problem, in this letter, a meta-based self-supervision learning method is introduced to find the optimal cache policy for minimizing the burst errors. In contrast to the traditional methods, the proposed one has better performance and learning efficiency. More particularly, the proposed can determine much proper cache policy by just training with a few samples of average fading duration of each channel. We demonstrate the effectiveness of the proposed method by numerical simulations which indicate that the proposed meta-learning method is superior to common existing methods in combat of the burst-error problem regarding convergence time and accuracy.

VC-PPQ: Privacy-Preserving Q-Learning Based Video Caching Optimization in Mobile Edge Networks

Mobile edge caching has great advantages in alleviating video traffic pressure and reducing transmission delay, which is considered as a hopeful solution to improve video resource utilization and achieve faster service response. However, due to insufficient and non-real time acquisition of the status between edge servers and mobile users, it is still difficult for existing solutions to improve the video quality of experience (VQoE) for mobile users. Besides, due to the mobility of users and malicious third-party attacks, users may suffer the risk of personal privacy (e.g. location, trajectory, preference, etc.) leakage when requesting video services that are scattered in different areas. In order to efficiently deal with these problems, this paper proposes a privacy-preserving Q-learning based video caching optimization framework in mobile edge networks (VC-PPQ), including a privacy scheme for user data and a dynamic video caching algorithm. Specifically, in the first task, the local differential privacy model is leveraged to protect the locations and preferences of users. Then, a data aggregation model is provided to make trade-off between aggregation accuracy and privacy protection. In the second task, we further obtain the video popularity in user's area by a computation service model, then formulate a caching objective function combined with transcoding technologies. Afterwards, Q-learning is adopted to obtain the goal of caching optimization, which can achieve low delay and high hit ratio. Finally, extensive simulations prove that our VC-PPQ framework delivers prominent performance advantages in terms of user privacy and caching services, which greatly improves the VQoE of users.

The Software Cache Optimization-Based Method for Decreasing Energy Consumption of Computational Clusters

Reducing the consumption of electricity by computing devices is currently an urgent task. Moreover, if earlier this problem belonged to the competence of hardware developers and the design of more cost-effective equipment, then more recently there has been an increased interest in this issue on the part of software developers. The issues of these studies are extensive. From energy efficiency issues of various programming languages to the development of energy-saving software for smartphones and other gadgets. However, to the best of our knowledge, no study has reported an analysis of the impact of cache optimizations on computing devices’ power consumption. Hence, this paper aims to provide an analysis of such impact on the software energy efficiency using the original software design procedure and computational experiments. The proposed Software Cache Optimization (SCO)-based Methodology was applied to one of the key linear algebra transformations. Experiments were carried out to determine software energy efficiency. RAPL (Running Average Power Limit) was used—an interface developed by Intel, which provides built-in counters of Central Processing Unit (CPU) energy consumption. Measurements have shown that optimized software versions reduce power consumption up to 4 times in relation to the basic transformation scheme. Experimental results confirm the effectiveness of the SCO-based Methodology used to reduce energy consumption and the applicability of this technique for software optimization.

RoofSplit: An edge computing framework with heterogeneous nodes collaboration considering optimal CNN model splitting

Intelligent applications based on AI have put much challenge on the edge of wireless networks, considering the heterogeneous characteristics and insufficient resources of the edge nodes. In this paper, we propose an edge computing framework named ’RoofSplit’ with the integration of communication, caching and computing ability. With this framework, deep learning models such as convolutional neural network (CNN) can be executed at the network edge with high efficiency. We first utilize Roofline theory and cut the one special complex CNN to several optimal sub-models. And then deploy the sub models on different edge nodes and process them in parallel. We also build a hardware testbed with the GPU and caching optimization to adapt to the RoofSplit framework. Finally, we make the experiments on the testbed. The results show that, compared with non-split scheme, RoofSplit can increase throughput by 4%–15% and reduce memory usage by 40%–60%. By using Redis caching, database is deemed as message queue in internal communication. The delay of processing one learning model is 40%–70% lower than that of Cloud-Only scheme, and more than 10% lower than that of Edge–Cloud collaboration scheme, which also reflects good real-time performance of the RoofSplit framework.

Fresh Caching of Dynamic Content Over the Wireless Edge

We introduce a framework and provably-efficient schemes for ‘fresh’ caching at the (front-end) local cache of content that is subject to ‘dynamic’ updates at the (back-end) database. We start by formulating the hard-cache-constrained problem for this setting, which quickly becomes intractable due to the limited cache. To bypass this challenge, we first propose a flexible time-based-eviction model to derive the average system cost function that measures the system’s cost due to the service of aging content in addition to the regular cache miss cost. Next, we solve the cache-unconstrained case, which reveals how the refresh dynamics and popularity of content affect optimal caching. Then, we extend our approach to a soft-cache-constrained version, where we can guarantee that the cache use is limited with arbitrarily high probability. The corresponding solution reveals the interesting insight that ‘whether to cache an item or not in the local cache?’ depends primarily on its popularity level and channel reliability, whereas ‘how long the cached item should be held in the cache before eviction?’ depends primarily on its refresh rate. Moreover, we investigate the cost-cache saving trade-offs and prove that substantial cache gains can be obtained while also asymptotically achieving the minimum cost as the database size grows.

Fundamental Structure of Optimal Cache Placement for Coded Caching With Nonuniform Demands

This paper studies the caching system of multiple cache-enabled users with random demands. Under nonuniform file popularity, we thoroughly characterize the optimal uncoded cache placement structure for the coded caching scheme (CCS). Formulating the cache placement as an optimization problem to minimize the average delivery rate, we identify the file group structure in the optimal solution. We show that, regardless of the file popularity distribution, there are <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">at most three file groups</i> in the optimal cache placement, where files within a group have the same cache placement. We further characterize the complete structure of the optimal cache placement and obtain the closed-form solution in each of the three file group structures. A simple algorithm is developed to obtain the final optimal cache placement by comparing a set of candidate closed-form solutions computed in parallel. We provide insight into the file groups formed by the optimal cache placement. The optimal placement solution also indicates that coding between file groups may be explored during delivery, in contrast to the existing suboptimal file grouping schemes. Using the file group structure in the optimal cache placement for the CCS, we propose a new information-theoretic converse bound for coded caching that is tighter than the existing best one. Moreover, we characterize the file subpacketization in the CCS with the optimal cache placement solution and show that the maximum subpacketization level in the worst case scales as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\mathcal{ O}}(2^{K}/\sqrt {K})$ </tex-math></inline-formula> for <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> users.

Multiagent Cooperative Caching Policy in Industrial Internet of Things

To handle the impact of the explosive growth of data traffic generated by smart industrial applications in the Industrial Internet of Things (IIoT) scenario, edge caching is commonly used in the IIoT to reduce the content access delay and to release the backhaul load. However, due to the mobility of IIoT devices and temporal dependence of popularity, traditional content caching policies, such as least frequently used (LFU) and least recently used (LRU), cannot cache the required contents accurately within the coverage of edge server. Therefore, by exploiting the moving trajectory of the IIoT devices and the temporal dependence of the content popularity, we propose a multiagent cooperative caching policy, in which each edge server acts as an agent to cooperatively learn the optimal caching decision, then each edge server caches the corresponding contents to reduce the content access delay. Especially, we first apply the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> -order Markov chain to predict the moving trajectory of the IIoT devices to get the IIoT device set within the coverage of each edge server. Second, we use long short-term memory (LSTM) to get the prior knowledge of the content requests by using the moving trajectory prediction results. Finally, we obtain the optimal caching decision by deep reinforcement learning to improve the Quality of Service (QoS) of the IIoT applications. Experimental results demonstrate that the proposed caching policy can efficiently improve the cache-hit ratio and content access delay.

Research on Discrete Artificial Bee Colony Cache Strategy of UAV Edge Network

Unmanned aerial vehicle edge networks (UENs) can reduce the cache load of the core network and improve system performance to provide users with efficient content services. However, the time-varying characteristics of content popularity in UENs lead to a low accuracy of popularity prediction, and the capacity limitations of wireless channel conditions lead to a lower cache hit rate than the rates of traditional fiber-optic-based cache strategies. Therefore, this paper proposes the discrete artificial bee colony cache strategy of UENs (DABCCSU). First, the information–dynamics–dissemination model of UENs (IDDMU) is established to deduce the coupling relationship between the channel capacity and the service probability in IDDMU. The influence of the service probability change on the content dissemination process is discussed, and the content popularity in UENs is predicted by the state iteration matrix. Then, the discrete artificial bee colony cache (DABCC) optimization algorithm is proposed. The action function of the artificial bee colony is designed as a random action based on the historical cache strategy. The discrete cache strategy is used as an optimization variable, and the popularity prediction result obtained by IDDMU is used to maximize the cache hit rate. DABCC provides the optimal cache strategy for the UENs, and effectively improves the cache hit rate. The simulation result shows that the accuracy of DABCCSU in content popularity prediction is more than 90%, which achieves a good prediction effect. In terms of cache performance, the average cache hit rate of DABCCSU is 91.62%, which is better than the 51.09% of the Least Recently Used (LRU) strategy, 89.27% of the Greedy Algorithm (GA) and 54.26% of Binary Particle Swarm Optimization (BPSO). In addition, the cache hit rate of DABCCSU under different cache capacities is better than that of LRU, GA, and BPSO, showing a relatively stable performance. It shows that DABCCSU can achieve excellent content popularity prediction, and it can also maximize the cache hit rate under limited communication resources and cache resources to provide UENs with the optimal content cache strategy, and provides users with high-quality content services.

Utility Optimization for Resource Allocation in Multi-Access Edge Network Slicing: A Twin-Actor Deep Deterministic Policy Gradient Approach

To achieve the service-oriented features of the 5G, network slicing aims to create logical virtual networks where multiple services are provided on a common physical infrastructure. The performance of network slicing depends on the intelligent management of multi-dimensional resources, which are exactly what multi-access edge computing (MEC) provides. This paper proposes joint optimization of communication, computing and caching (3C) resources in multi-access edge network slicing. The optimization objective of the two-level resource allocation problem is to maximize the utility obtained by mobile virtual network operators while ensuring the quality of service (QoS). The deep reinforcement learning (DRL) approach is employed which enables the resource allocation scheme to intelligently adapt to the dynamic environment. Specifically, we propose a novel DRL approach named twin-actor deep deterministic policy gradient (twin-actor DDPG). Since the action space is continuous, the DDPG is adopted where the actor generates the deterministic policy while the critic evaluates the policy and guides the actor to obtain the optimal policy. A novel twin-actor structure is put forward to replace the actor of the DDPG, thus the slice-level action and user-level action can be generated respectively. The convergence and effectiveness of the proposed DRL based algorithm is are verified by numerical simulation.