Edge Caching for Mobile Networks
2: Princeton University, New Jersey, USA
Caching enables the storage of Internet-based content, including web objects, videos and software updates. When web objects are downloaded from the Internet or across wide-area network (WAN) links, edge caching stores them at the edge of the network. Content can also be proactively cached at the edge based on its predicted popularity. When subsequent requests come for cached material, the content is quickly delivered from edge caching, without the need to download the data again over the WAN. The result is the ability to help save bandwidth, particularly at times of peak network load, increase content delivery, and provide users with a faster and better network experience. In this comprehensive edited book, the editors and authors introduce edge caching in mobile networks from a fundamental perspective and discuss its role in saving bandwidth and reducing latency over wireless channels. Many physical layer models and techniques, including interference alignment and beamforming are considered, as well as recent advances on intelligent and proactive communication systems capable of recommending content to users to improve quality of experience and spectral efficiency. The book provides systematic and thorough coverage of edge caching for mobile networks for an audience of researchers, engineers and scientists from academia and industry working in the fields of information and communication technology, data science and AI.
Inspec keywords: radio networks; mobile radio; mobile computing; telecommunication traffic; cache storage
Other keywords: telecommunication traffic; mobile networks; radio networks; recommender systems; cellular radio; mobile computing; resource allocation; edge caching; telecommunication computing; probability; mobile radio; cache storage
Subjects: General electrical engineering topics; Mobile, ubiquitous and pervasive computing; Radio links and equipment; File organisation; General and management topics
- Book DOI: 10.1049/PBTE096E
- Chapter DOI: 10.1049/PBTE096E
- ISBN: 9781839531224
- e-ISBN: 9781839531231
- Page count: 789
- Format: PDF
-
Front Matter
- + Show details - Hide details
-
p.
(1)
-
1 Introduction
- + Show details - Hide details
-
p.
1
–16
(16)
Caching has attracted significant attention recently because it holds the promise of scaling the service capability of mobile networks. Modern mobile networks are capable of achieving data rates of Gbps, yet they may still fail to meet the predicted bandwidth requirements of future mobile applications. A recent report from Cisco forecasts that mobile data traffic will grow to 77.49 EB per month in 2022. In theory, a human brain may process up to 100 T bits per second. Even though this does not necessarily reflect the ultimate bandwidth required by a human being, a huge gap between bandwidth demand and provisioning may still exist in the near future. Unfortunately, on-demand transmission which dominates current mobile network architectures has almost achieved its performance limits revealed by Shannon in 1948. Because of this and other limitations it is time to envision paradigm-shifting mobile network architectures for the sixth generation (6G) of mobile networks. In this chapter, we shall present a brief overview of the technical content of this book. In particular, this book mainly covers five major topics belonging to edge caching for mobile networks. They are fundamental limits of caching, asymptotic analysis of caching in large mobile networks, resource allocation for edge caching, caching with time-varying popularity and recommendation systems, and some typical applications of edge caching.
-
2 Physical layer schemes for coded caching
- + Show details - Hide details
-
p.
17
–42
(26)
In this chapter, we focus on the elegant and information-theoretic near-optimal coded caching scheme introduced by Maddah-Ali and Niesen (for brevity referred to as MAN). The MAN scheme was originally proposed for an ideal shared link network, where all users receive the same error-free broadcast transmission from the content server.
-
3 Cache-aided content delivery over noisy broadcast channels
- + Show details - Hide details
-
p.
43
–73
(31)
The focus of this chapter is on the benefits of coded caching over wireless broadcast channels (BCs) taking into account the physical characteristics of the wireless medium. The seminal coded caching technique by Maddah-Ali and Niesen, considers delivery over an error-free rate limited BC from the server to the users. However, in practice the delivery takes place over a noisy BC, and users may experience different quality channels. Hence, a more practical delivery technique must take the varying channel quality of the users into account. We will first consider a memoryless packet erasure BC from the server to the users, where the users have arbitrary erasure probabilities during the delivery phase. Equal-size caches are assigned to the users with the worse channel conditions in order to compensate for the weaker channel quality. The trade-off between the rate of the files in the library that can be delivered to the users reliably and the cache size available at the users is provided. Next, a Gaussian BC from the server to the users is considered during the delivery phase, while the placement phase is carried out without any channel state information (CSI). The goal is to characterize the trade-off between the transmit power at the server and the cache size available at the users.
-
4 Multi-antenna cache-aided channels
- + Show details - Hide details
-
p.
75
–98
(24)
This chapter discusses coded caching in multi-antenna systems and argues that when caching and multi-antenna delivery are treated jointly, multi-antenna systems can benefit from a powerful performance boost.
-
5 Caching for interference networks: a separation architecture
- + Show details - Hide details
-
p.
99
–118
(20)
The focus of this chapter is cache-enabled wireless networks, which are typically deployed for the last communication hop to the end user. Content distribution architectures designed for the wired Internet, including CCN/ICN, cannot be directly translated to this wireless setting. This is because the wireless last hop is a communication bottleneck link, and because the wireless cellular base station typically has limited storage. Therefore, last-hop content delivery schemes need to be designed cognizant of the wireless architecture being deployed.
-
6 Scalable delivery of correlated video content over cache-aided broadcast networks
- + Show details - Hide details
-
p.
119
–154
(36)
The increasingly dynamic, unpredictable, and personalized nature of the content that users consume challenges the efficiency of existing caching-based solutions in which only exact content reuse is explored. This calls for novel content delivery schemes that account for the additional gains that can be obtained from the joint compression of correlated content distributed throughout the network, such as updated versions of dynamic data. This chapter reviews state-of-the-art studies on the fundamental limits of cache-aided communication systems for the delivery of correlated content. Two scenarios are considered: (i) a static setting, in which the same correlated content library is used during both the caching and delivery phases and (ii) a dynamic setting, in which an updated version of the content library may become available during the delivery phase.
-
7 Rate-distortion theory for caching
- + Show details - Hide details
-
p.
155
–188
(34)
In this chapter, we consider a single-server K-user centralized caching setup with a library of correlated files. In the first communication phase, the placement phase, the library prefetches and stores content in the users' cache memories. Later, each user demands one specific file from the library and the demands are revealed to the server and all the users in the system. During the second communication phase, the delivery phase, the server sends a common delivery message that is observed by all the users in the system and allows the users to reconstruct, using the contents in their local cache memory, their requested file with the desired distortion. The main focus of this chapter is on the minimum rate of this delivery message given a set of distortion criteria and cache sizes at the users. We call this quantity the rate-distortion-memory (RDM) function of the system.
-
8 Secrecy of edge caching
- + Show details - Hide details
-
p.
189
–219
(31)
Wireless networks are naturally vulnerable to a variety of security threats. The common thread is to complement developments in any aspect of wireless technologies with security-preserving mechanisms. To this end, we study in this chapter various types of confidentiality concerns and adversarial attacks in caching networks. The underlying model assumptions and network topology are decidedly chosen to abstract the security concern in interest in its most basic formulation. The defining problem, in most of the considered settings, is to study the fundamental trade-off between the cache-memory sizes in the network and the delivery load such that the user requests and the desired security requirements are satisfied. Characterizing such trade-offs is approached by constructing achievability algorithms and deriving impossibility bounds. A particular emphasis is given to a setting in which a powerful strategic adversary optimizes its tapping capability over the two phases of communication in a cache-aided system, i.e., cache placement and delivery.
-
9 Scaling laws for cache-aided device-to-device networks for wireless video
- + Show details - Hide details
-
p.
221
–254
(34)
In this chapter, we will first review scaling laws for "standard" (non-caching) D2D networks, as well as other caching approaches that do not make use of D2D communications (see Section 9.1.2). Section 9.2 summarizes the system model that underlies most scaling law investigations. This is followed by a review of scaling laws for single-hop and multihop networks in Sections 9.3 and 9.4, respectively. Next, we discuss the scaling laws when coded caching is combined with D2D communications in Section 9.5. A discussion of how the scaling laws relate to practical implementation rounds off the chapter.
-
10 Caching increases the capacity of wireless networks
- + Show details - Hide details
-
p.
255
–280
(26)
Content Caching at the network edge, e.g., base stations and mobile devices, is a promising solution to meet the demand of massive mobile traffic growth. In this chapter, we focus on edge caching in a wireless ad hoc network without any infrastructure support, and study how content caching at mobile devices can help improve the network capacity. In wireless ad hoc networks, due to the interference between concurrent transmissions, the per-node capacity generally decreases with the increasing number of nodes in the network. Caching can help improve the network capacity, as it shortens the content transmission distance and reduces the communication interference. However, the fundamental performance limits of caching in wireless ad hoc networks have rarely been studied in an analytical manner. In this chapter, we first illustrate the benefit of caching under uniform content popularity, i.e., the capacity of wireless ad hoc networks with caching can remain constant even as the number of nodes in the network increases. Then, we generalize the results, evaluate how the distribution of the content popularity affects the per-node capacity, and derive different capacity scaling laws based on the skewness of the content popularity. Our results suggest that for wireless networks with caching, when contents have skewed popularity, increasing the number of nodes monotonically increases the per-node capacity.
-
11 Cache-induced hierarchical cooperation in wireless D2D networks
- + Show details - Hide details
-
p.
281
–337
(57)
In this chapter, we combine wireless device caching and hierarchical cooperation to significantly improve the capacity of wireless D2D networks. Specifically, we propose a cache-induced hierarchical cooperation scheme where the network is abstracted as a tree graph with each virtual node in the graph representing a cluster of nodes in the network, and the content files are cached at different levels of the tree graph according to their popularities.
-
12 Wireless device-to-device caching networks with distributed MIMO and hierarchical cooperations
- + Show details - Hide details
-
p.
339
–370
(32)
This chapter first categorizes and summarizes wireless caching on its capability of raising the wireless network throughput in centralized and decentralized network topology. More precisely, what is the throughput scaling law of the wireless caching networks, i.e., how does the network throughput scale with the caching size, the library size, and the number of nodes in different wireless networks? Then, a novel caching scheme for device-to-device (D2D) networks is presented and the throughput scaling law is derived, which shows that the average aggregate throughput of the network scales almost linearly with the number of nodes. Both analytical and numerical results show that the proposed scheme outperforms existing ones when the local cache size is limited.
-
13 Randomised geographic caching and its applications in wireless networks
- + Show details - Hide details
-
p.
371
–404
(34)
The randomised (or probabilistic) geographic caching is a proactive content placement strategy that has attracted a lot of attention, because it can simplify a great deal cache-management problems at the wireless edge. It diversifies content placement over caches and applies to scenarios where a request can be possibly served by multiple cache memories. Its simplicity and strength is due to randomisation. It allows one to formulate continuous optimisation problems for content placement over large homogeneous geographic areas. These can be solved to optimality by standard convex methods and can even provide closed-form solutions for specific cases. This way the algorithmic obstacles from NP-hardness are avoided and optimal solutions can be derived with low computational cost. Randomised caching has a large spectrum of applications in real-world wireless problems, including FemtoCaching, multi-tier networks, device-to-device (D2D) communications, mobility, mm-wave, security, UAVs and more. In this chapter, we will formally present the main policy with its applications in various wireless scenarios. We will further introduce some very useful extensions related to unequal file sizes and content placement with neighbourhood dependence.
-
14 Pricing-aided proactive caching for mobile services
- + Show details - Hide details
-
p.
405
–438
(34)
In this chapter, we study the potential of harnessing "proactive caching" at end-users, supported with pricing incentives, to maximize the profit of wireless content providers (CPs). CP consistently encounters time varying, yet partially predictable, demand characteristics. The disparate demand levels throughout the course of the day yield excessive service cost in the peak hours that substantially hurt the reaped profit. With the CP's ability to track and statistically predict future requests of its users, we propose to enable proactive caching of the peak hour demand during the off-peak times. Thus network traffic will be smoothed out, while end-users' activity patterns are undisturbed. In addition, the CP is able to assign personalized pricing policies that strike a best balance between enhancing certainty about the future demand for optimal proactive caching, and maximizing the revenue collected from end-users. These proactive caching decisions accompanied with smart data pricing offer a win-win situation to both the CP and the end-users. In this chapter, we propose a joint proactive caching and content pricing scheme that archives such a win-win situation. Comparing the proposed system's performance with the baseline scenario of the existing practice of no-proactive caching, we show that the CP attains profit gain that grows with the number of users, at least, as the first derivative of the cost function. Moreover, end-users that receive proactive caching services make strictly positive savings. Thus, we essentially demonstrate the win-win situation to be reaped by exploiting consistent users' activity. In addition, we reveal significant (potentially unbounded) gains and low-complexity means of the smart pricing paradigm for proactive caching.
-
15 Mobile edge caching: an optimal auction approach
- + Show details - Hide details
-
p.
439
–462
(24)
With an explosive growth of wireless data, the sheer volume of mobile traffic is challenging the capacity of current wireless systems. To tackle this challenge, mobile edge caching has emerged as a promising paradigm, in which service providers (SPs) prefetch some popular content items in advance and cache them locally at the network edge. When requested, those locally cached content items can be directly delivered to users with low latency, thus, alleviating the traffic load over backhaul channels during peak hours and enhancing the quality-of-experience (QoE) of users simultaneously. Owing to the limited available cache space, it makes sense for the SP to cache the most profitable content items. Nevertheless, users' true valuations of content items are their private knowledge, which is unknown to the SP in general. This information asymmetry poses a significant challenge for effective caching at the SP side. Further, the cached content items can be delivered with different levels of quality, which needs to be chosen judiciously to balance delivery costs and user satisfaction. To tackle these difficulties, we propose an optimal auction mechanism from the perspective of the SP. In the auction, the SP determines the cache space allocation over content items and user payments based on the users' (possibly untruthful) reports of their valuations so that the SP's expected revenue is maximized. The advocated mechanism is designed to elicit true valuations from the users (incentive compatibility) and to incentivize user participation (individual rationality). In addition, we devise a computationally efficient method for calculating the optimal cache space allocation and user payments. We further examine the optimal choice of the content delivery quality for the case with a large number of users and derive a closed-form solution to compute the optimal delivery quality. Finally, extensive simulations are implemented to evaluate the performance of the proposed optimal auction mechanism, and the impact of various model parameters is highlighted to obtain engineering insights into the content caching problem.
-
16 Contract theory-based approach for caching
- + Show details - Hide details
-
p.
463
–491
(29)
With the dramatically growing demands on video entertainment and social connections, wireless multimedia communications have played an important role in future cellular networks. Unfortunately, the capacity of current networks has not been able to keep a similar pace with the tremendous growth of cellular traffic. At the same time, it is pointed out that the next generation of mobile communication systems will have to meet the following requirements: high throughput, high access amount, high data rate, low power consumption and low latency. In order to fulfil these requirements, researchers have begun conceiving 5G cellular networks and are focusing on key techniques, such as millimetre wave technology, massive multiple-input multiple-output and super density heterogeneous networks. However, these techniques may need to intensively change the hard-ware equipment and network protocols, imposing high cost and complexity to current communication systems. It is urgent to find a solution which is more economical and convenient.
-
17 Mean-field game-theoretic edge caching
- + Show details - Hide details
-
p.
493
–517
(25)
Mobile networks are envisaged to be extremely densified in 5G and beyond to cope with the ever-growing user demand. Edge caching is a key enabler of such an ultra-dense network (UDN), through which popular content is prefetched at each small base station (SBS) and downloaded with low latency while alleviating the significant backhaul congestion between a data server and a large number of SBSs. Focusing on this, in this chapter we study the content caching strategy of an ultra-dense edge caching network (UDCN). Optimizing the content caching of a UDCN is a crucial yet challenging problem. Owing to the sheer amount of SBSs, even a small misprediction of user demand may result in a large amount of useless data cached in capacity-limited storages. Furthermore, the variance of interference is high due to short inter-SBS distances, making it difficult to evaluate cached data downloading rates, which is essential in optimizing the caching file sizes. To resolve these problems, we first present a spatio-temporal user demand model in continuous time, in which the long-term and short-term content popularity variations at a specific location are modelled using the Chinese restaurant process (CRP) and the Ornstein-Uhlenbeck (OU) process, respectively. Based on this, we aim to develop a scalable and distributed edge caching algorithm by leveraging the mean-field game (MFG) theory.
-
18 Caching with time-varying popularity profiles
- + Show details - Hide details
-
p.
519
–536
(18)
This chapter considers the problem of content caching at the small-cell base stations (sBSs) in a heterogeneous wireless network. A cost function called the "offloading loss," which measures the fraction of the requested files that are not available in the sBS caches, is used as a metric for evaluating the performance of the proposed caching scheme. In contrast to the previous approaches that consider time-invariant and perfectly known popularity profiles, caching with non-stationary and statistically dependent popularity profiles (assumed unknown, and hence, estimated) is studied from a learning-theoretic perspective. A high probability bound on the offloading loss difference referred to as a probably approximately correct (PAC) result is derived. Here, the offloading loss difference refers to the error between the estimated and the optimal offloading loss. The difference is shown to be a function of the Rademacher complexity, the β-mixing coefficient, the number of time slots, and a measure of discrepancy between the estimated and true popularity profiles. Using insights from this bound, a practical cache update algorithm is proposed. Simulation results are presented to show its superiority over periodic updates. This chapter also presents caching performance analyses for Bernoulli and Poisson request models.
-
19 Reinforcement learning for caching with space-time popularity dynamics
- + Show details - Hide details
-
p.
537
–563
(27)
With the tremendous growth of data traffic over wired and wireless networks along with the increasing number of rich-media applications, caching is envisioned to play a critical role in next-generation networks. To intelligently prefetch and store contents, a cache node should be able to learn what and when to cache. Considering the geographical and temporal content popularity dynamics, the limited available storage at cache nodes, as well as the interactive influence of caching decisions in networked caching settings, developing effective caching policies is practically challenging. In response to these challenges, this chapter presents a versatile reinforcement learning-based approach for near-optimal caching policy design, in both single-node and network caching settings under dynamic space-time popularities. The policies presented here are complemented using a set of numerical tests, which showcase the merits of the presented approach relative to several standard caching policies.
-
20 Push-based wireless converged networks with user caching
- + Show details - Hide details
-
p.
565
–579
(15)
With rapid development of the emerging information and communication technologies, e.g., the mobile Internet, the Internet of Things, and social networks, mobile applications continue to be enriched, and the demand for mobile data is showing an explosive growth trend. It is widely expected that the data traffic of mobile devices in 2030 will be 1,000 times the traffic in 2020. Nowadays, the mobile data is mainly delivered through the mobile cellular network, e.g., 4G/5G. However, recent studies have shown that the fast growing traffics would soon become a significant burden on the cellular infrastructure and cause severe congestion in the near future,
-
21 On the joint optimization of content caching and recommendations
- + Show details - Hide details
-
p.
581
–613
(33)
Content caching has been experiencing revived interest within the context of current and next generation wireless networks. It brings content closer to users, decreasing the aggregate delivery costs and service delays for network providers. Recommender systems have become integral components of content provision sites. They offer personalized recommendations, increasing the individual user satisfaction and engagement with the content provider's platform. Traditionally, caching and recommendation decisions are taken separately. However, there is a recent persistent trend where both network and content providers tend to deploy their own content delivery solutions. In light of this, we explore how the phenomenally conflicting objectives of content caching and recommendation can be jointly addressed. In this chapter we approach recommender systems as network traffic engineering tools that actively shape content demand to serve both user- and network-centric performance objectives. We introduce a model that captures the coupling between caching decisions and issued recommendations. Based on experimental evidence, we describe the impact of recommendations on user content requests and present a systematic way of engineering the user recommendations.
-
22 Caching with recommendation and reinforcement learning
- + Show details - Hide details
-
p.
615
–642
(28)
The benefit of proactive caching at mobile devices hinges on the accurate prediction of user demands and mobility, which, however, is hard to achieve due to the uncertain user behaviour. In this chapter, we show how to leverage personalized content recommendation to reduce the uncertainty in user demands for increasing the caching gain at mobile devices. Specifically, a joint content pushing and recommendation problem is formulated for maximizing the net profit of the mobile network operator. To cope with the challenges in modeling and learning user behaviour, a reinforcement learning (RL) framework is introduced for solving the problem. Since the joint problem encounters the curse of dimensionality due to very large action and state spaces, it is decomposed into two sub-problems where two agents (namely a recommendation agent and a pushing agent) having different goals cooperate. Then, the sub-problems are solved by using the double deep-Q network with duelling architecture. Our simulation results show that the learned recommendation and pushing policies increase the net profit remarkably when compared with the baseline policies.
-
23 Social networking-driven caching
- + Show details - Hide details
-
p.
643
–674
(32)
As a promising technology in future communication systems and networks, wireless caching is envisioned to revolutionize the wireless resource management in 5G and B5G. By storing content files closer to the user end, caching can effectively alleviate the load on the content server and reduce the content retrieving delay. In particular, distributed storage/caching system (DSS/DCS) has gradually attracted the attention of academia and industry by storing content on multiple cache servers in a distributed manner to improve system storage efficiency, information security, and scalability. To further enhance system robustness and reliability, the social networks should also be taken into consideration to evaluate the relationship among caching entities for comprehensive caching scheme design. We first introduce the basic concept of caching, and then the significance of considering social networks in caching will be presented.
-
24 Caching for secure social networks
- + Show details - Hide details
-
p.
675
–698
(24)
MSNs (mobile social networks) have become one of the most important networking paradigms in future wireless mobile networks. Recent advances of caching, as well as computing and communications, can have significant impacts on the performance of MSNs. Specifically in this chapter, recent advances in mobile edge computing, in-network caching, and D2D communications in MSNs are studied. In addition, the knowledge of social relationships in these new paradigms is investigated to improve the security and efficiency of MSNs, where a social trust scheme is present with both direct observation using Bayesian inference and indirect observation using the Dempster-Shafer theory. Deep Q-learning approach is used to study this complicated system. Simulation results were presented to show the effectiveness of the proposed scheme.
-
25 Caching for CRAN and UAVs
- + Show details - Hide details
-
p.
699
–727
(29)
In cloud radio access networks (CRANs), mobile users are serviced by a massive number of remote radio heads (RRHs) that are acted as distributed antennas and controlled by cloud-based baseband units (BBUs) via wired or wireless fronthaul links. To improve spectral efficiency, cloud-based cooperative signal processing techniques can be executed centrally at the BBUs. However, despite the ability of CRAN systems to run such complex signal processing functions centrally, their performance remains limited by the capacity of the fronthaul (users to BBU) and backhaul (BBUs to core) links. Indeed, given the massive nature of a CRAN, relying on fiber fronthaul and backhaul links may be infeasible. Consequently, capacity-limited wireless backhaul and fronthaul connections are being studied for CRANs. To overcome these limitations, one can make use of content caching techniques in which users can directly obtain contents stored at the cloud or RRHs. However, deploying caching strategies in a CRAN environment faces many challenges that include optimized cache placement, cached content update, and cached content delivery. In this chapter, we first study the application of echo state network to predict the users' mobility and content request distribution for determining optimal cached contents in CRAN. Then, we study the deployment of cache-enabled UAVs to service ground mobile users.
-
26 Dynamic computation offloading and resource allocation in cache-assisted mobile edge computing systems
- + Show details - Hide details
-
p.
729
–750
(22)
Mobile edge computing (MEC) is one of the most promising techniques for next-generation wireless communication systems. In this chapter, we study the problem of dynamic caching, computation offloading, and resource allocation in cache-assisted multi-user MEC systems with stochastic task arrivals. There are multiple computationally intensive tasks in the system, and each mobile user (MU) needs to execute a task either locally or remotely in an MEC server by offloading the task data. Popular tasks can be cached in the MEC server to avoid duplicates in offloading. The cached contents can be either obtained through user offloading or fetched from a remote cloud. The objective is to minimize the long-term average of a cost function, which is defined as a weighted sum of energy consumption, delay, and cache contents' fetching costs. The weighting coefficients associated with the different metrics in the objective function can be adjusted to balance the trade-off among them. The optimum design is performed with respect to four decision parameters: whether to cache a given task, whether to offload a given uncached task, how much transmission power should be used during offloading, and how much MEC resources to be allocated for executing a task. We propose to solve the problem by developing a dynamic scheduling policy based on deep reinforcement learning (DRL) with the deep deterministic policy gradient (DDPG) method. Simulation results demonstrate that the proposed algorithm outperforms other existing strategies such as deep Q-network (DQN).
-
Back Matter
- + Show details - Hide details
-
p.
(1)