Smart manufacturing describes the new manufacturing intelligence applied to modern information technologies such as the Internet of Things, cloud computing, and artificial intelligence in the manufacturing process (Tao et al. 2018a, b). Data from various sources is becoming integrated into manufacturing intelligence to improve manufacturing in various ways. A typical example is the wide deployment of sensors in manufacturing to monitor and provide real-time manufacturing data such as temperature, humidity, speed, vibration, and acidity for better decision-making and control of the manufacturing process (Li et al. 2015). Therefore, multiple sensors of different modalities are needed in distributed locations. Wired sensor networks are extensively adopted, but the cost for their installation, testing, maintenance, and shutdown are quite high. In many cases, a wireless sensor network (WSN) is a more attractive alternative because it does not require any fixed infrastructure and can be applied over distributed areas where cabling is costly (Akyildiz et al. 2002). A WSN is composed of sensor nodes and sink nodes. Sensor nodes are responsible for collecting and forwarding data. Sink nodes either analyze data locally or forward data to a base station (Magaia et al. 2015). Each sensor node consists of small devices for sensing, processing, transceiving, and power, and it is able to communicate with other sensor nodes or directly with a base station. A WSN can be regarded as a self-supporting unit, with which unattended operation can be realized (Stankovic 2008). It is mostly applicable where power supplies and cabling are difficult, or in hostile environments that people cannot enter. With the technological developments in cloud computing and Internet of Things, WSNs are becoming even more widely deployed. They have also been used for machine health monitoring (Tiwari et al. 2007), data center monitoring (Wang et al. 2013), and data logging (Saleem et al. 2014) in various industries. Moreover, a WSN can be used for environmental monitoring (Hart and Martinez 2006), remote health care monitoring (Malasinghe et al. 2017), and other monitoring scenarios. Overall, its application in cloud computing and Internet of things and has significant social and economic benefit (Tao et al. 2014, 2011). WSNs have a key difference with the wired sensor network in that the sensor nodes in a WSN rely solely on their own battery, and thus have limited power resources. Moreover, their computing power and storage resources are also limited. As a consequence, reducing the energy consumption of each sensor node is a critical issue for WSNs (Carrabs et al. 2015). Ant colony optimization (ACO) has been used in WSNs to identify shortest paths, and thus reduce the energy consumed by a network. However, the ACO is prone to falling into local optima and converges slowly. We hence propose an improved ACO (IACO) that can be used to construct the sensor node transfer function and pheromone update rule, and adaptively construct a data route using the characteristics of a dynamic network. The major contributions of this paper are listed below. 1. The IACO consumes less energy. 2. The transmission delay is reduced using the IACO. 3. Fewer transmission packets are lost when using the IACO. This paper is organized as follows. Section 2 analyzes the current WSN routing algorithms and their weakness. In Sect. 3, we describe our model and its notation. In Sect. 4, a detailed description of the proposed routing IACO algorithm is given. In Sect. 5, simulation results comparing the proposed algorithm with several other algorithms are presented to show the effectiveness of the proposed method.