Global video content distribution networks (CDNs) serve a significant fraction of the entire Internet traffic. Effective caching at the edge is vital for the feasibility of these CDNs, which can otherwise incur substantial costs and overloads in the Internet. We analyze the challenges and requirements for content caching on the servers of these CDNs which cannot be addressed by standard solutions. We design multiple algorithms for this problem: a LRU-based baseline to address the requirements; a flexible ingress-efficient algorithm; an offline cache aware of future requests (greedy) to estimate the maximum efficiency we can expect from any online algorithm; an optimal offline cache (for limited scales); and an adaptive ingress control algorithm for reducing the server's peak upstream traffic. We use anonymized actual data from a global video CDN to evaluate the algorithms and draw conclusions on their suitability for different settings.

Video Content Distribution Networks (CDNs) serve a significant fraction of the Internet traffic through a global network of cache servers. In a planet-scale CDN with millions of videos, cache servers only consider their own request patterns for managing their content. We analyze how, in the absence of cooperative caching, the knowledge of requests in remote serving locations can lead to better caching decisions overall and can reduce serving costs. We call this practice cache coordination. Our analyses in this paper are based on actual video workload data from a global CDN. We analyze the spatial correlation of video popularities worldwide, the effectiveness and feasibility of cache coordination, and its scalability: from a city to across countries.

Cloud based video delivery platforms serve a significant fraction of the entire Internet traffic and are continuously expanding with the growing demand. We study the provisioning of the server cluster to deploy at each location in such systems. We optimize the right server count, peering bandwidth, and server configuration as the disk size and the necessary SSD and/or RAM caches to sustain the intensive I/O load. Our analyses are based on actual server traces from a global content delivery platform. Our optimization captures the interaction of cache layers in each server, the interplay between egress/disk capacity and network bandwidth, storage read/write constraints and storage prices.

We study delivering delay-sensitive data to a group of receivers with minimum latency. This latency consists of the time that the data spends in overlay links as well as the delay incurred at each overlay node, which has to send out a piece of data several times over a finite-capacity network connection. The latter part is a significant portion of the total delay as we show in the paper, yet it is often ignored or only partially addressed by previous multicast algorithms. We analyze the actual delay in multicast trees and consider building trees with minimum-average and minimum-maximum delay. We show the NP-hardness of these problems and prove that they cannot be approximated in polynomial time to within any reasonable approximation ratio. We then present a set of algorithms to build minimum-delay multicast trees that cover a wide range of application requirements-min-average and min-max delay, for different scales, real-time requirements, and session characteristics. We conduct comprehensive experiments on different real-world datasets, using various overlay network models. The results confirm that our algorithms can achieve much lower delays (up to 60% less) and up to orders-of-magnitude faster running times (i.e., supporting larger scales) than previous related approaches.

Planet-scale video Content Delivery Networks (CDNs) deliver a significant fraction of the entire Internet traffic. Effective caching at the edge is vital for the feasibility of these CDNs, which can otherwise incur significant monetary costs and resource overloads in the Internet.

Delivering delay-sensitive data to a group of receivers with minimum latency is a fundamental problem for various distributed applications. In this paper, we study multicast routing with minimum end-to-end delay to the receivers. The delay to each receiver in a multicast tree consist of the time that the data spends in overlay links as well as the latency incurred at each overlay node, which has to send out a piece of data several times over a finite-capacity network connection. The latter portion of the delay, which is proportional to the degree of nodes in the tree, can be a significant portion of the total delay as we show in the paper. Yet, it is often ignored or only partially addressed by previous multicast algorithms. We formulate the actual delay to the receivers in a multicast tree and consider minimizing the average and the maximum delay in the tree. We show the NP-hardness of these problems and prove that they cannot be approximated in polynomial time to within any reasonable approximation ratio. We then present a number of efficient algorithms to build a multicast tree in which the average or the maximum delay is minimized. These algorithms cover a wide range of overlay sizes for both versions of our problem. The effectiveness of our algorithms is demonstrated through comprehensive experiments on different real-world datasets, and using various overlay network models. The results confirm that our algorithms can achieve much lower delays (up to 60% less) and up to orders of magnitude faster running times (i.e., supporting larger scales) than previous minimum-delay multicast approaches.

To improve rendered video quality and serve more receivers, peer-to-peer (P2P) video-on-demand streaming systems usually deploy seed servers. These servers complement the limited upload capacity offered by peers. In this paper, we are interested in optimally managing the capacity of seed servers, especially when scalable video streams are served to peers. Scalable video streams are encoded in multiple layers to support heterogeneous receivers. We show that the problem of optimally allocating the seeding capacity to serve scalable streams to peers is NP-complete. We then propose an approximation algorithm to solve it. Using the proposed allocation algorithm, we develop an analytical model to study the performance of P2P video-on-demand streaming systems and to manage their resources. The analysis also provides an upper bound on the maximum number of peers that can be admitted to the system in flash crowd scenarios. We validate our analysis by comparing its results to those obtained from simulations. Our analytical model can be used by administrators of P2P streaming systems to estimate the performance and video quality rendered to users under various network, peer, and video characteristics.

Peer-to-peer (P2P) systems generate a major fraction of the current Internet traffic, and they significantly increase the load on ISP networks and the cost of running and connecting customer networks (e.g., universities and companies) to the Internet. To mitigate these negative impacts, many previous works in the literature have proposed caching of P2P traffic, but very few (if any) have considered designing a caching system to actually do it. This paper demonstrates that caching P2P traffic is more complex than caching other Internet traffic, and it needs several new algorithms and storage systems. Then, the paper presents the design and evaluation of a complete, running, proxy cache for P2P traffic, called pCache. pCache transparently intercepts and serves traffic from different P2P systems. A new storage system is proposed and implemented in pCache. This storage system is optimized for storing P2P traffic, and it is shown to outperform other storage systems. In addition, a new algorithm to infer the information required to store and serve P2P traffic by the cache is proposed. Furthermore, extensive experiments to evaluate all aspects of pCache using actual implementation and real P2P traffic are presented.

In recent years, peer-to-peer (P2P) and peer-assisted streaming have emerged as promising models for low-cost multimedia distribution to large scale user communities. In this paper, we study streaming of scalable video streams over these systems. Scalable video streams are composed of multiple layers and can easily be adapted according to the characteristics and needs of receivers. Thus, they can efficiently support a wide spectrum of heterogeneous peers participating in a P2P streaming system. We present an analytical model for forecasting the long-term behavior of a P2P streaming system with scalable video streams. Our analysis takes as inputs the characteristics of a dynamic P2P streaming system and the video streams. It then analytically computes the expected throughput of the streaming system and the expected video quality delivered to peers. The analysis also provides an upper bound on the maximum number of peers that can be admitted to the system at once (i.e., in flash crowd scenarios), while ensuring a certain video quality. We present a general analysis framework that can be customized to various practical P2P streaming systems with different characteristics. Then, we show the detailed analysis of a typical P2P streaming system and we explain how other systems can be analyzed using our model. We validate our analysis by comparing its results to those obtained from simulations, which confirm the accuracy of our analysis. Our analysis and simulations enable administrators of P2P streaming systems to predict the throughput and the video quality that can be delivered to users.

The large prevalence of multimedia systems in recent years makes the security of multimedia communications an important and critical issue. We study the problem of securing the delivery of scalable video streams so that receivers can ensure the authenticity of the video content. Our focus is on recent scalable video coding (SVC) techniques, such as H.264/SVC, which can provide three scalability types at the same time: temporal, spatial, and visual quality. This three-dimensional scalability offers a great flexibility that enables customizing video streams for a wide range of heterogeneous receivers and network conditions. This flexibility, however, is not supported by current stream authentication schemes in the literature. We propose an efficient and secure authentication scheme that accounts for the full scalability of video streams, and enables verification of all possible substreams that can be extracted from the original stream. In addition, we propose an algorithm for minimizing the amount of authentication information that need to be attached to streams. The proposed authentication scheme supports end-to-end authentication, in which any third-party entity involved in the content delivery process, such as stream adaptation proxies and caches, does not have to understand the authentication mechanism. Our simulation study with real video traces shows that the proposed authentication scheme is robust against packet losses, incurs low computational cost for receivers, has short delay, and adds low communication overhead. Finally, we implement the proposed authentication scheme as an open source library called svcAuth, which can be used as a transparent add-on by any multimedia streaming application.

With the rapid increase in the demand for multimedia services, securing the delivery of multimedia content has become an important issue. Accordingly, the problem of multimedia stream authentication has received considerable attention by previous research and various solutions have been proposed. However, these solutions have not been rigorously analyzed and contrasted to each other, and thus their relative suitability for different streaming environments is not clear. This article presents comprehensive analysis and comparison among different schemes proposed in the literature to authenticate multimedia streams. Authentication schemes for nonscalable and scalable multimedia streams are analyzed. To conduct this analysis, we define five important performance metrics, which are computation cost, communication overhead, receiver buffer size, delay, and tolerance to packet losses. We derive analytic formulas for these metrics for all considered authentication schemes to numerically analyze their performance. In addition, we implement all schemes in a simulator to study and compare their performance in different environments. The parameters for the simulator are carefully chosen to mimic realistic settings. We draw several conclusions on the advantages and disadvantages of each scheme. We extend our analysis to authentication techniques for scalable streams. We pay careful attention to the flexibility of scalable streams and analyze its impacts on the authentication schemes. Our analysis and comparison reveal the merits and shortcomings of each scheme, provide guidelines on choosing the most appropriate scheme for a given multimedia streaming application, and could stimulate designing new authentication schemes or improving existing ones. For example, our detailed analysis has led us to design a new authentication scheme that combines the best features of two previous schemes.

We study streaming of scalable videos over peer-to-peer (P2P) networks. We focus on efficient management of seed servers resources, which need to be deployed in the network to make up for the limited upload capacity of peers in order to deliver higher quality video streams. These servers have finite serving capacity and are often loaded with a volume of requests larger than their capacity. We formulate the problem of allocating this capacity for optimally serving scalable videos. We show that this problem is NP-complete, and propose two approximation algorithms to solve it. The first one allocates seeding resources for serving peers based on dynamic programming, and is more suitable for small seeding capacities (les 10 Mbps). The second algorithm follows a greedy approach and is more efficient for larger capacities. We evaluate the proposed algorithms analytically and in a simulated P2P streaming system. The results confirm the efficiency and near-optimality of the proposed algorithms, and show that higher-quality videos are delivered to peers if our algorithms are employed for allocating seed servers.

We investigate the problem of securing the delivery of scalable video streams so that receivers can ensure the authenticity (originality and integrity) of the video. Our focus is on recent scalable video coding techniques, e.g., H.264/SVC, that can provide three scalability types at the same time: temporal, spatial, and quality (or PSNR). This three-dimensional scalability offers a great flexibility that enables customizing video streams for a wide range of heterogeneous receivers and network conditions. This flexibility, however, is not supported by current stream authentication schemes in the literature. We propose an efficient authentication scheme that accounts for the full scalability of video streams: it enables verification of all possible substreams that can be extracted and decoded from the original stream. Our evaluation study shows that the proposed authentication scheme is robust against packet losses, adds low communication and computation overheads, and is suitable for live streaming systems as it has short delay.