报告包括电子商务方面的重要观点、政策与环境、网购平台、行业应用、电商物流、电商金融等研究。

软件定义网络(software defined network,简称SDN)通过集中式的控制器提高了网络的可编程性,成为近年来网络领域非常热门的话题.以Openflow网络为代表的软件定义网络将逻辑控制与数据转发相隔离,为网络虚拟化技术提供了良好的平台.集中式的抽象与控制使得SDN虚拟化框架的处理效率成为主要瓶颈.现有的SDN虚拟化框架由于缺乏对细粒度并行的支持,为编程人员充分利用多核/众核资源、控制更大规模的网络带来了极大的挑战.为了提高SDN虚拟化框架的处理效率,提出一种新的SDN虚拟化编程框架,通过新颖的API和运行时,在框架内部支持细粒度的并行处理.该框架通过对网络中流和网络资源进行抽象,使开发人员可以直接通过划分流空间来定义不同的虚拟网络,利用无锁的编程方式对共享的网络资源和流进行操作.实验结果表明,该框架在逻辑控制的执行效率方面具有良好的可扩展性,可以创建出更大规模的虚拟网络,并对其进行更为复杂的控制.

随着物联网技术的不断发展,传感器网络得到了广泛的应用并成为信息技术领域重要的基础设施.尤其是传感网络提供的实时感知信息,为许多智能应用提供了充分的信息支持和必要的决策依据.然而,由于智能应用的实时感知信息需求通常无法转化为简单的查询请求与传感器底层查询接口准确匹配,因此,基于物联网的智能决策常常无法准确获取到决策相关的实时信息.针对此问题,提出一个基于语义覆盖网的物联网信息资源描述、推理和应用模型,并以多智能体系统决策支持为应用基础,研究了新型物联网环境下的多智能体决策信息支持技术.该技术以基于多智能体系统的团队导向规划的任务分解方法为核心,将复杂任务分解为若干简单子任务,并基于本体推理方法把子任务执行时需要的决策信息转化为精确、完备的传感器信息查询,从而实现从物联网中准确定位具体的传感器并获取相应感知信息的实时决策信息支持机制.

已有传感器网络中,过滤机制只能在转发过程中过滤虚假数据而无法过滤重复数据,且无法防范协同攻击.提出了一种基于单向哈希链的过滤方案HFS.在HFS中,节点在部署后将密钥和初始哈希值预分发给部分中间节点存储,每个数据包附带t个MAC和新鲜哈希值,转发节点同时对数据包中检测节点之间相对位置关系的合法性、MAC和哈希值的正确性以及哈希值的新鲜性进行验证.理论分析及仿真实验结果表明,HFS可同时过滤传感器网络中的虚假数据和重复数据,并能有效对抗协同攻击.

网络虚拟化环境下的跨域虚拟网络映射是指当物理网络由多个自治域构成时,以最小化虚拟网络映射开销为目标,将虚拟网络请求恰当地划分为多个虚拟子网请求,并分别指派给相应自治域以完成映射.资源匹配和虚拟网络划分是跨域虚拟网络映射中的两个关键阶段.然而,现有的资源匹配算法无法支持精确的数值属性匹配,也无法满足虚拟网络用户对表达多样化映射约束的需求,故实用性不高.此外,虚拟网络划分属于NP问题,目前也缺乏高效的求解方法.针对上述两个阶段中存在的问题,分别提出了基于OWL及SWRL的资源匹配算法和基于遗传算法的虚拟网络划分算法.理论分析证明了该方法的正确性.仿真实验从效率、性能及稳定性方面验证了该方法的有效性.

Recent investigations into the fundamental limits of mobile ad hoc networks have produced a physical layer method for approaching their capacity. This strategy, known as interference alignment, requires cooperation, rather than competition, among the transceivers in the network. Under ideal circumstances, every user can achieve reliable communication at rates approaching one half of the interference-free capacity. Unfortunately, prior research on interference alignment makes many assumptions that make it difficult to realize interference alignment in practice. This report summarizes the PI's work on realizing interference alignment in multiple-input multiple-output (MIMO) communication channels. The emphasis is on networks where clusters of users cooperate together, since this provides a good balance between network overhead and performance. One contribution was to devise a way to add new users to an existing cluster of interference aligned users. Different algorithms were devised as a function of the number of antennas in the network. Another contribution was to analyze interference alignment in a network with random clusters of users. The cases where interference alignment was preferred to other simpler communication strategies was characterized. Another contribution was to determine the impact of distributing the antennas. It was found that distributed antennas lead to higher performance with interference alignment, and new algorithms were designed to achieve that performance. A final contribution was to prototype an interference alignment network. A distributed prototype was constructed with over-the-air synchronization and feedback, demonstrating the viability of interference alignment for small networks in practice.

Navigation requires knowledge of current location and a planned destination. This is true with manned vehicles and unmanned vehicles. There are many ways to acquire the current location, including global positioning system (GPS), triangulation, radar, and dead reckoning. Today GPS is the most reliable and accurate navigation technique when there is a clear, unobstructed view of the satellite constellation. Various sensors can be used to perform indoor navigation; however, when the vehicle is autonomous the sensors need to provide the exact location to the system. This research determined if using a template replay strategy has the same RNR performance as using an analog noise source. Using the template replay approach, each RNR node has a priori knowledge about the transmitted waveforms of other nodes. The analysis here revealed that modifications do not significantly alter RNR functionality. The analysis revealed that even at SNIR equal to 0 dB, there are no parameters that can be reliably extracted other than transmitted signal bandwidth and transmitted template length; the transmitted message length was able to be extracted because the message was repeated over and over. If the message was not replayed the analysis showed that there would be no ability to extract parameters. Finally, by using the RNR to transmit digitally generated templates, digital communication is possible and the symbol error rate (SER) is traceable to simulated SER.