Air Option 1: Technology Translation - Network Deduplication for Smartphones and Tablets

此PFI：AIR技术转换项目专注于转换一种用于智能手机和平板电脑的高级网络重复数据删除(Dedup)技术，称为智能缓存，以满足无线用户数量爆炸性增长对更好的无线网络性能的日益增长的需求。转换后的智能缓存解决方案包括一套将传统重复数据删除扩展到移动环境的策略，并具有以下独特功能：(1)非对称缓存，允许删除重复数据目标(dd-dst)向dd-src反馈其缓存的内容，否则删除重复数据消除源(dd-src)可能不知道这些内容。(2)PreActing，当后者通过低成本网络接入连接时，允许DD-DST机会性地预取内容，以便当这些内容实际通过高成本网络访问时，可以将成本降至最低。这两个变种的研究贡献在于为应对移动环境中存在的独特挑战而专门构建的底层算法。初步的基于跟踪的性能分析表明，智能缓存将蜂窝网络上的流量消耗降低了85%到96%，从而显著提高了性能。非对称缓存本身比对称缓存提高了125%。该项目通过开发一个可行的智能缓存解决方案原型来实现这一目标，从而可以解决该技术商业化过程中的一个重要的可信度差距。虽然生产智能缓存解决方案的方法有多种，但我们选择构建一个可供无线网络运营商和第三方解决方案提供商采用的原型。-智能缓存服务器：该项目涉及构建一个智能缓存服务器，作为一种独立的网络设备，可以部署在无线运营商网络内或第三方解决方案提供商的代理环境中。构建这样的设备允许在目标环境中随时部署，而无需担心计算资源的可用性。虽然产品形式将是设备的形式，但实施将完全在Linux操作系统上运行的软件中实现。因此，如果需要，可以轻松地将原型移植到目标环境中的虚拟机上。-智能缓存客户端：该项目涉及为Android和iOS设备构建智能缓存客户端。到目前为止，这两个移动平台主导着当今的移动设备版图。虽然Windows Mobile和Blackberry RIM等操作系统是其他相关平台，但在成功完成拟议的原型工作后，可以为这些其他平台开发或移植解决方案。合作伙伴关系与佐治亚理工学院风险投资实验室合作，在无线市场动态和商业化和产品化的其他方面提供指导，因为它们与智能缓存沿着一条可能导致竞争商业现实的道路的潜力有关。智能缓存的潜在经济影响是深远的，这将有助于美国在电信行业的竞争力。无线服务提供商可以采用智能缓存来立即减少昂贵频谱的使用。或者，第三方解决方案提供商可以将智能缓存作为一种成本和性能优化服务直接提供给最终用户。该项目还做出了以下更广泛的影响贡献：(I)行业：预计85%的企业知识型员工将在未来几年内拥有智能手机。智能缓存方面的工作将直接惠及广大企业，因为更好的移动接入将提高员工的工作效率和客户参与度。(2)教育：此外，我们认为，这项努力通过学生积极参与研究和教学，对本科生和研究生都产生了积极的影响。

This PFI: AIR Technology Translation project focuses on translating an advanced network deduplication (dedup) technology for smartphones and tablets, called Smart-Cache, to fill the ever increasing need for better wireless network performance driven by the explosion in the number of wireless users.The translated smart-cache solution consists of a suite of strategies to extend traditional dedup to mobile environments, and has the following unique features: (1) Asymmetric caching, which allows the dedup destination (dd-dst) to feedback to the dd-src the contents of its cache that otherwise might not be known to the dedup source (dd-src). (2) PreActing, that allows the dd-dst to opportunistically prefetch content when the latter is connected through a low-cost network access so that when such content is actually accessed through a high-cost network, the costs may be minimized. The research contributions of these two variations lie in the underlying algorithms that are purpose-built to handle the unique challenges present in mobile environments. Preliminary trace-based performance analysis shows that smart-cache significantly improves the performance by reducing traffic consumption over cellular networks to the tune of 85% to 96%. Asymmetric caching by itself shows an improvement of 125% over symmetric caching.The project accomplishes this goal by developing a viable prototype of the smart-cache solution so that an important credibility gap in the commercialization process for the technology may be addressed. While there are several avenues for productizing the smart-cache solution, we choose to build a prototype that could be adopted by both wireless network operators and third party solution providers. At a high level, we propose to build the following:- Smart-cache server: The project involves the building of a smart-cache server as a stand-alone network appliance that can be deployed within the wireless operators network or within a proxy environment of a third-party solutions provider. Building such an appliance allows for ready deployment in a target environment without concerns about the availability of computational resources. While the product form will be that of an appliance, the implementation will entirely be in software running on a Linux operating system. Hence, the prototype can easily be ported onto virtual machines in the target environment if necessary. - Smart-cache clients: The project involves the building of a smart-cache client for both Android and iOS devices. These two mobile platforms by far dominate the mobile device landscape today. While operating systems such as Windows Mobile and Blackberry RIM are other relevant platforms, solutions can later be developed or ported for these other platforms after the proposed prototyping effort is successfully accomplished.The partnership engages the Georgia Tech VentureLab to provide guidance in wireless market dynamics and other aspects of commercialization and productization as they pertain to the potential to translate smart-cache along a path that may result in a competitive commercial reality.The potential economic impact of smart-cache is profound, which will contribute to the U.S. competitiveness in telecommunication industry. A wireless service provider could adopt smart-cache to immediately reduce the usage of expensive spectrum. Alternatively, a third party solutions provider could offer smart-cache as a cost and performance optimization service directly to the end-user. The project also makes the following broader impact contributions: (i) Industry: 85% of enterprise knowledge workers are expected to have smartphones within the next few years. The work on smart-cache will directly benefit the broad swathe of enterprises as better mobile access will improve employee productivity and customer engagement. (ii) Education: In addition we see the effort positively impacting both undergraduate and graduate level students through aggressive involvement of the students in both research and teaching.