The emergence of mobile service composition meets the current needs for real-time eCommerce. However, the requirements for eCommerce, such as safety and timeliness, are becoming increasingly strict. Thus, the cloud-edge hybrid computing model has been introduced to accelerate information processing, especially in a mobile scenario. However, the mobile environment is characterized by limited resource storage and users who frequently move, and these characteristics strongly affect the reliability of service composition running in this environment. Consequently, applications are likely to fail if inappropriate services are invoked. To ensure that the composite service can operate normally, traditional dynamic reconfiguration methods tend to focus on cloud services scheduling. Unfortunately, most of these approaches cannot support timely responses to dynamic changes. In this article, the cloud-edge based dynamic reconfiguration to service workflow for mobile eCommerce environments is proposed. First, the service quality concept is extended. Specifically, the value and cost attributes of a service are considered. The value attribute is used to assess the stability of the service for some time to come, and the cost attribute is the cost of a service invocation. Second, a long short-term memory (LSTM) neural network is used to predict the stability of services, which is related to the calculation of the value attribute. Then, in view of the limited available equipment resources, a method for calculating the cost of calling a service is introduced. Third, candidate services are selected by considering both service stability and the cost of service invocation, thus yielding a dynamic reconfiguration scheme that is more suitable for the cloud-edge environment. Finally, a series of comparative experiments were carried out, and the experimental results prove that the method proposed in this article offers higher stability, less energy consumption, and more accurate service prediction.

中文翻译：

---

面向移动电子商务环境的基于云边缘的动态重新配置服务工作流程

移动服务组合的出现满足了当前对实时电子商务的需求。然而，电子商务对安全性和时效性等要求越来越严格。因此，引入了云边混合计算模型来加速信息处理，尤其是在移动场景中。然而，移动环境的特点是资源存储有限，用户频繁移动，这些特点强烈影响了在该环境中运行的服务组合的可靠性。因此，如果调用不适当的服务，应用程序可能会失败。为了保证组合服务能够正常运行，传统的动态重配置方法往往侧重于云服务调度。很遗憾，这些方法中的大多数不能支持对动态变化的及时反应。在本文中，提出了基于云边缘的移动电子商务环境服务工作流动态重新配置。一是服务质量理念的延伸。具体来说，要考虑服务的价值和成本属性。value 属性用于评估服务在未来一段时间内的稳定性，cost 属性是服务调用的成本。其次，使用长短期记忆（LSTM）神经网络来预测服务的稳定性，这与价值属性的计算有关。然后，针对可用设备资源有限的情况，介绍一种计算调用服务成本的方法。第三，通过考虑服务稳定性和服务调用成本来选择候选服务，从而产生更适合云边缘环境的动态重配置方案。最后进行了一系列对比实验，实验结果证明本文提出的方法稳定性更高，能耗更低，服务预测更准确。