A transport protocol for content-centric networking with explicit congestion control

摘要: Content-centric networking (CCN) adopts a receiver-driven, hop-by-hop transport approach that facilitates in-network caching, which in turn leads to multiple sources and paths for transferring content. In such case, keeping single round trip time (RTT) estimator multi-path flow is insufficient as each path may experience different times. To solve this problem, it has been proposed use RTT estimators predict network condition. paper, we examine an alternative CHoPCoP, utilizes explicit congestion control cope with the multiple-source, multiple-path situation. Protocol design innovations of CHoPCoP include random early marking (REM) scheme explicitly signals congestion, per-hop fair share Interest shaping algorithm (FISP) receiver method (RIC) regulate rates at routers respectively. We have implemented on ORBIT testbed conducted experiments under various traffic settings. The evaluation shows viable can effectively deal multipath environment. I. INTRODUCTION During last decade, content retrieval dominated Internet usage. address challenges posed retrieval, content-centric (12), (15) proposed. Being significant shift philosophy, CCN centered named instead host addresses. Routing towards based name address, data initiated by issuing receiver. Com- pared application-layer overlay solutions Content Distribution Networks (CDN) Peer-to-peer systems (P2P), holds promise providing more efficient cost- effective solution dissemination. CCN's unique characteristics introduce new chal- lenges underlying protocol. First, naturally since needs issue first order request Data chunk. Second, transfer desired transport, because con- tent files be cached along route improve throughput. Moreover, specific often widely disseminated network, - i.e., one chunk originates from source A, while next might originate B. Such multi-source/multi-path makes estimation value fall short. These features are sufficiently distinct traditional end-to-end host-based model called for. Recently protocol received