SBIR Phase II: A Robust Caller-ID Alternative for Securing Telephony Based Transactions

该项目的创新来自于新颖的呼叫音频分析技术，该技术可用于创建电话呼叫源的指纹。这种指纹可以揭示有关呼叫来源的宝贵信息，包括呼叫设备的类型（座机、voip或移动电话）、其地理位置以及呼叫音频在到达被叫方之前可能被传输的网络。为了利用这种指纹实时检测潜在的欺诈性呼叫，该提案计划扩展呼叫源地理位置的识别，并创建一个实时执行音频分析的主动呼叫分析器。第二阶段项目的重点是探索主动调用分析器的设计选项；包括准确性、可扩展性和及时性的权衡及其在呼叫中心基础设施中的集成。该电话分析器还将用于建立电话诈骗情报服务，该服务将主动检测用于诈骗的电话号码。这样的服务，包括与合作伙伴和客户共享情报的机制，可以帮助保护电话通道免受各种攻击。这个项目的更广泛的影响和商业化潜力可以很容易地从观察中看出，电话欺诈已经是多个部门的一个严重问题，包括银行、医疗保健甚至执法部门。此外，由于欺诈性呼叫已经造成了相当大的经济损失，呼叫中心的客户代理正在询问多种基于知识的问题来验证呼叫者的身份。这会导致更高的呼叫处理成本，也会降低客户体验。一个成功的主动呼叫分析解决方案可以自动生成呼叫者身份验证的风险评分，这将产生广泛的影响，因为整个服务部门都依赖呼叫中心进行客户联系，它可以降低成本并改善客户体验。该项目还将使Pindrop在组织更广泛的社区中发挥领导作用，发起一个类似于反网络钓鱼工作组的电话反欺诈联盟。由这样一个小组提供的思想领导对于定义电话安全挑战和解决这些挑战的方法是必要的。同样，这将确保项目对多个行业产生广泛影响。

The innovation in this project comes from the novel call audio analysis techniques that can be used to create a fingerprint of the source of a telephone call. Such fingerprints can reveal valuable information about the call source, including the type of the calling device (landline, Voice-over IP or mobile), its geographical location and the networks over which the call audio may have been transported prior to it reaching the called party. To detect potentially fraudulent calls in real-time with such fingerprints, this proposal plans to extend the identification of the geography of a call source and the creation of an active call analyzer that performs audio analysis in real-time. The focus of this Phase II project is on exploring design options for an active call analyzer; including accuracy, scalability and timeliness tradeoffs and its integration in call center infrastructures. The call analyzer will also be used to build a phone fraud intelligence service that proactively detects phone numbers used for committing fraud. Such a service, including mechanisms for sharing of intelligence with partners and customers, can help secure the telephony channel from a variety of attacks.The broader impact and commercialization potential of this project can be seen readily from the observation that phone fraud is already a serious problem for multiple sectors, including banking, healthcare and even law enforcement. Also, because fraudulent calls are already responsible for considerable financial loss, customer agents in call centers are asking multiple knowledge-based questions to authenticate a caller. This leads to higher costs for call handling and also degrades customer experience. A successful active call analyzer solution that can automatically generate a risk score for caller authentication will have broad impact because the entire service sector relies on call centers for customer contact and it could reduce costs and improve customer experience. The project will also enable Pindrop to play a leadership role in organizing the broader community to launch a phone anti-fraud alliance similar to the anti-phishing working group. The thought leadership provided by such a group will be necessary to define the telephony security challenges and approaches for addressing them. Again, this will ensure broad impact of the project across several industries.