Joint forwarding and caching with latency awareness in information-centric networking

详细信息    查看全文

• 作者：Giovanna Carofiglio^a ; ^{gcarofig@cisco.com" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Lé ; once Mekinda ; ^b ; ^{leonce.mekinda@orange.com" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Luca Muscariello^a ; ^{lumuscar@cisco.com" class="auth_mail" title="E-mail the corresponding author}Author Vitae
• 关键词：ICN ; Caching ; Adaptative forwarding ; Network performance analysis
• 刊名：Computer Networks
• 出版年：2016
• 出版时间：9 December 2016
• 年：2016
• 卷：110
• 期：Complete
• 页码：133-153
• 全文大小：2904 K

文摘

Research on 5G has recently promoted latency minimization from a critical network optimization criterion to an architectural cornerstone. Information-Centric Networking (ICN) appears a promising candidate technology for building an agile communication model that reduces latency via a fully distributed and adaptive delivery approach coupling in-network caching and forwarding. In the paper, we investigate theoretically and empirically the role of latency awareness on multipath ICN delivery performance and analyze FOCAL, an approach combining novel caching and forwarding strategies to jointly reduce end-user experienced latency with no network signaling nor coordination between routers.

FOCAL gathers a latency-proportional probabilistic caching policy, with a load-aware dynamic forwarding strategy, that preferentially routes popular content requests through a single path (set of caches), while globally achieving minimum network load and user content delivery time, thus delay minimization. By means of ICN simulations, we assess the advantages of FOCAL over existing alternatives given by the combinations of known caching policies and forwarding strategies. FOCAL drastically improves end-user delivery performance as it reduces by up to 60% the mean and variance of content delivery time. It also results in a faster convergence to these figures, even under the varying network conditions induced by non-stationary content popularity distributions.