Utility-driven caching opened up a new design opportunity for caching algorithms by modeling the admission and eviction control as a utility maximization process with essential support for service differentiation. Nevertheless, there is still to go in terms of adaptability to changing environment. Slow convergence to an optimal state may degrade actual user-experienced utility, which gets even worse in non-stationary scenarios where cache control should be adaptive to time-varying content request traffic. This paper proposes to exploit deep reinforcement learning (DRL) to enhance the adaptability of utility-driven time-to-live (TTL)-based caching. Employing DRL with long short-term memory helps a caching agent learn how it adapts to the temporal correlation of content popularities to shorten the transient-state before the optimal steady-state. In addition, we elaborately design the state and action spaces of DRL to overcome the curse of dimensionality, which is one of the most frequently raised issues in machine learning-based approaches. Experimental results show that policies trained by DRL can outperform the conventional utility-driven caching algorithm under some non-stationary environments where content request traffic changes rapidly.

In serverless-enabled edge computing, container startup delay is one of the most critical issues because it violates some quality-of-service (QoS) requirements such as the ultra-low latency response times. Caching critical packages for the container can mitigate the startup delay associated with container instantiation. However, caches consume the memory resource that is highly limited at edge nodes. It means that the package cache must be carefully managed in serverless-enabled edge computing. This paper proposes a deep reinforcement learning (DRL)-based caching algorithm, which efficiently caches critical and popular packages with per-function response time QoS in hierarchical edge clouds. By conducting multi-agent reinforcement learning (MARL) for the caching agents of on-premise edge nodes in conjunction with a global reward that considers both cache hit and QoS violation numbers, the caching agents can be driven to cooperate with each other. The results of simulation demonstrate that the proposed DRL-based caching policy can improve QoS awareness more effectively than baselines. Compared with the LRU and LFU, the rate of violation fell by 18 and 27 percent, respectively.

As the data size on smart things continue to grow, it becomes inefficient to transfer the massive raw data to centralized cloud datacenters for the analysis of raw data and the deployment of innovative and novel services into various areas such as smart cities, etc. As the concept of Fog Computing, the structure of distributed intelligent infrastructure can address those problems by providing elastic resources and services to end users with smart devices at the edge of network. Hence, through the functions of analytics and flexible control based on the Software‐Defined Network (SDN), it is necessary to provide the SDN‐driven management intelligence, which can maximize network utilization subject to constraints on service priority and resources fairness by differentiating service flows according to the quality requirements. This paper proposes the SDN controller for intelligent inter‐datacenter (inter‐DC) cloud networking, which enables optimal traffic loads distribution across the distributed datacenters with the global inter‐DC view of link states and flow statistics by applying the network‐aware flow scheduling on a policy‐driven inter‐DC traffic control layer. The proposed intelligent inter‐DC controller performs a global resources scheduling mechanism that uses different weights according to the priority of service class in allocating bandwidths based on multi‐layers networking interconnecting the distributed datacenters.

The performance of Functions-as-a-Service (FaaS) would be significantly improved by organizing cloud servers into a hierarchical distributed architecture, resulting in low-latency access and faster response when compared to centralized cloud. However, the distributed organization introduces a new type of decision making problem for placing and executing functions to a specific cloud server. In order to handle the problem, we extended a well-known cloud computing simulator, CloudSim. The extended CloudSim enables users to define FaaS functions with various characteristics and service level objectives (SLOs), place them across geo-distributed cloud servers, and evaluate per-function performance. Proof-of-Concept (PoC) evaluation results show the potential of our CloudSim extension in terms of execution efficiency and simulation reality.

Traffic monitoring is the essential capability for large-scale enterprises, data centers, service providers, and network operators to ensure reliability, availability, fault assurance, and security of their underlying network resources. Currently, most monitoring solutions are standalone dedicated ones. Major drawbacks of these dedicated standalone appliances per-feature are high-cost, lack of flexibility, slow install time and difficulty of maintenance. Network Function Virtualization (NFV) provides an attractive alternative to cope with such limitations by controlling both CAPEX and OPEX. Network traffic monitoring function virtualization brings not only CAPEX/OPEX advantages but also introduces some challenges such as ensuring scalability and performance of single or distributed multiple virtual monitoring functions, utilization of virtual functions, and flexibility and easiness of virtual functions lifecycle management. To address such challenges, we propose a novel architecture and proof-of-concept implementation of a software-defined Virtual TrAffic Monitoring function with resource Auto-Scaling capability built over a multi-core whitebox server (VTAMAS) in this paper. It virtualizes monitoring functions with the capability of auto-scaling resource of virtual functions for efficient resource utilization especially.

The paper investigates theoretical issues in applying the universal swarming technique to efficient content distribution. In a swarming session, a file is distributed to all the receivers by having all the nodes in the session exchange file chunks. By universal swarming, not only all the nodes in the session, but also some nodes outside the session may participate in the chunk exchange to improve the distribution performance. We present a universal swarming model where the chunks are distributed along different Steiner trees rooted at the source and covering all the receivers. We assume chunks arrive dynamically at the sources and focus on finding stable universal swarming algorithms. To achieve the throughput region, universal swarming usually involves a tree-selection subproblem of finding a min-cost Steiner tree, which is NP-hard. We propose a universal swarming scheme that employs an approximate tree-selection algorithm. We show that it achieves network stability for a reduced throughput region, where the reduction ratio is no more than the approximation ratio of the tree-selection algorithm. We propose a second universal swarming scheme that employs a randomized tree-selection algorithm. It achieves the throughput region, but with a weaker stability result. Comprehensive simulation results support the stability analysis of the algorithms. The proposed schemes and their variants are expected to be useful for infrastructure-based content distribution networks with massive content and relatively stable network environment.

A distinct trend has emerged that the Internet is used to transport data on a more and more massive scale. Capacity shortage in the backbone networks has become a genuine possibility, which will be more serious with fiber-based access. The problem addressed in this paper is how to conduct massive content distribution efficiently in the future network environment, where the capacity limitation can equally be at the core or the edge. We propose a novel technique as a main content transport mechanism to achieve efficient network resource utilization. The technique uses multiple trees for distributing different file pieces, which at the heart is a version of swarming. In this paper, we formulate an optimization problem for determining an optimal set of distribution trees as well as the rate of distribution on each tree under bandwidth limitation at arbitrary places in the network. The optimal solution can be found by a distributed algorithm. The results of the paper not only provide stand-alone solutions to the massive content distribution problem, but should also help the understanding of existing distribution techniques such as BitTorrent or FastReplica.

A distinct trend has emerged that the Internet is used to transport data on a more and more massive scale. Capacity shortage in the backbone networks has become a genuine possibility, which will be more serious with fiber-based access. The problem addressed in this paper is how to conduct massive content distribution efficiently in the future network environment where the capacity limitation can equally be at the core or the edge. We propose a novel peer-to-peer technique as a main content transport mechanism to achieve efficient network resource utilization. The technique uses multiple trees for distributing different file pieces, which at the heart is a version of swarming. In this paper, we formulate an optimization problem for determining an optimal set of distribution trees as well as the rate of distribution on each tree under bandwidth limitation at arbitrary places in the network. The optimal solution can be found by a distributed algorithm. The results of the paper not only provide stand-alone solutions to the massive content distribution problem, but should also help the understanding of existing distribution techniques such as BitTorrent or FastReplica.

We propose a residential gateway architecture for IPTV service that has QoS management functionality. The proposed residential gateway contains a multicast packet classifier and a priority regenerator that can handle multiple IPTV traffics concurrently. We also implement an IP tuner for an IP settop box that can extract MPEG streams from IP packets and has the same role as a legacy tuner of a digital TV settop box. In order to provide IPTV service with QoS in a home network, we design an IPTV channel zapping manager that communicates coincidently with a home network QoS manager.

There are many efforts to adapt streaming services to wireless mobile networks. It is hard to assure the QoS of the streaming services in wireless mobile networks because a terminal moves continuously and causes handoff procedures. We focus on delay factors which occur whenever a terminal moves in the wireless mobile environment. It is assumed that a mobile terminal's movement is based on the Gauss-Markov model with the varying velocity affected by its previous velocity. Using this model, we can predict the mobility pattern of a moving terminal and the delay of streaming services caused from handoff procedures during its moving. To compensate the streaming service delay, the paper proposes a dynamic rate control scheme for streaming services based on the next-handoff prediction using the Gauss-Markov model, which is the predictive mobility model based on the varying velocity and the handoff occurrence rates according to the time.