In this project, we aim at developing automatic procedures to decentralize (decompose and distribute) and optimize Internet of Things (IoT) aware Business Processes (BPs).

Two companies will provide real use cases that we will use to apply and validate project results in real context.

The results of this project are as relevant as the increase use of IoT in business processes, since it contributes to solve scale problems related to, mainly, central processing, message communication, and battery consumption.
On the one hand, we will use predefined information we will get from the process definition, for instance. In addition, we will also use information about execution logs.

The information that process definitions provide will be use to:

  1. Adapt the WSNs communication behaviour according to business process requirements. In previous work, we introduce quality of information and access cost requirements of IoT resources into BMPN process definitions. To reflect these requirements into WSN behaviour, network middleware of WSNs needs to be flexible so, for instance, it can send urgent and non-urgent messages through different interfaces (such as 3G cellular network or WIFI). This flexibility comes at a cost of memory and storage usage,
    a challenge regarding physical object constraints.
  2. Decentralize IoT aware business processes through WSNs as well as intermediate network nodes. In previous works, we decentralize business processes through WSNs just considering control flow dependencies. In this project, we will include data dependencies as well as WSNs capabilities. In addition, we also consider intermediate network nodes as execution targets. This way, we have to deal with additional constraints and additional execution targets.
  3. Complementary with the decentralization approach, we will use information about execution logs to optimize the distribution of business processes through WSNs and intermediate network nodes. This requires collecting events information from WSNs without penalizing communications, handling event logs to associate them with process instances so we have useful process logs, correlating process logs, and finally using this information for optimization. We will apply process mining techniques and we will formulate this optimization problem as ana daptation of a quadratic assignment problem (QAP), with additional constraints, considering two weights in data flow edges.

Despite the wide range of applications that can benefit from the results of this project, we will use two smart cities’ applications:

  • One for automatic irrigation;
  • One in the area of waste management, leveraging our collaboration with two companies.

Our team brings together researchers with previous work on the range of topics that need to be addressed: sensor networks, sensor programming, heuristics, optimization, business processes, IoT aware business processes, and automatic irrigation.
The team is complemented with two companies with knowledge and experience in the areas of our project use cases.