Overview
Wireless sensor networks are an information acquisition and processing technology that integrates the logical information world with the physical world. Wireless sensor networks have broad application prospects across many domains.
Cloud computing is a flexible IT resource organization and delivery model. It supports distributed storage and parallel processing, and its data processing framework handles much of the data locally without requiring large-scale remote transfers.
Cloud Sensors
Cloud sensors integrate measurement, network transmission, and cloud components. They combine sensing and networking components so that sensor data is uploaded directly to a cloud platform, and users access the data from the cloud, reducing intermediate transmission and processing circuitry.
Advantages
1. Simple to use: after power-up, users can read sensor measurements directly from a computer or a mobile device.
2. Cloud processing: cloud sensors are connected in real time. The cloud platform can perform large-scale data processing on real-time sensor data and can also feed results back to the sensors. A direct application is online dynamic calibration of sensors, using large datasets to reduce local measurement errors.
3. Easier secondary development: users can avoid complex cloud deployment and reduce development cost.
Cloud-based Wireless Sensor Network
1. Architecture
The architecture for a cloud-based wireless sensor network places a group of special nodes within the WSN area, called cloud nodes. Cloud nodes have richer resources than ordinary sensors and perform two functions. First, they form part of the cloud. Second, they communicate with sensors to collect nearby sensor data; these nodes are also called point sinks. Sensors send sensed data to a point sink using multi-hop routing; the point sink then stores the data in the cloud. Sensors that send data to the same point sink form a partition. For sensors, the whole cloud appears as a virtual sink.
2. Partitioned Sensor Organization
Sensors belong to a partition according to a rule. For example, a sensor can join the partition represented by the nearest point sink. Sensors in a partition can be homogeneous or heterogeneous; they form a local WSN that is independent of other partitions. Local WSNs are connected through the cloud. Sensors in a partition can be organized as a flat structure or a hierarchical structure.
In a multi-layer WMSN structure, the same type of sensors are organized at the same tier, forming a WSN. Lower-tier WSNs connect to higher tiers via central nodes, and higher-tier central nodes forward data for the lower tiers. Higher-tier central nodes can also process forwarded data and schedule activities of sensors in that tier, such as sleep or wake. Compared with keeping all sensors active, this reduces sensor energy consumption, but it increases the complexity of software design and maintenance. In addition, central nodes handle and forward large volumes of data, so their energy drains faster.
In the cloud-based architecture, another organization approach partitions sensors by type. Each sensor type forms its own WSN and sends data to the same point sink (if same-type sensors cannot communicate directly, different-type sensors can act as gateways). Each sensor type forms a logical WSN that may overlap physically with others; these WSNs connect to the same point sink. Data processing software is deployed in the cloud rather than only at the connected point sinks. After collecting and processing various sensed data, the cloud can schedule sensors in each independent WSN via the point sinks. Compared with single-layer cluster-based structures, this approach yields smaller independent WSN sizes within a partition, higher transmission efficiency, and global visibility of control information.
3. Cloud Organization
Cloud node deployment must consider: a) how cloud nodes are deployed in the WSN area; b) how cloud nodes communicate with each other; c) how nodes organize into a cloud; d) how data collected by point sinks is stored and processed in the cloud.
The placement of cloud nodes or point sinks depends on the sensor distribution in the WSN. Ideally, point sink positions minimize overall network data transmission operations; complex algorithms can be used for placement. However, due to WSN dynamics such as topology changes, route changes, and variable data generation patterns, ideal placement is difficult. When sensors are uniformly distributed, a simple approach is to make each point sink responsible for approximately equal numbers of sensors.
Cloud nodes must communicate with each other to form the cloud. Cloud nodes should be reasonably deployed within the WSN area and operate in a manageable environment, so network infrastructure may be present. If wired networks are available, nodes can use VPNs for secure communication. If mobile networks such as 2.5G or 3G are available, cloud nodes can interconnect via those networks. In infrastructure-free WSN deployments, self-organizing wireless networks are the only option; for example, cloud nodes can be configured with an IEEE 802.11 stack to form an ad hoc network.
Sensor Data Upload to the Cloud
Wireless standards have changed significantly, enabling sensors to synchronize large amounts of data to the cloud and retrieve it as needed. The potential benefits of the Internet of Things are widely recognized. Various solutions exist; in practice, the more useful and cost-effective approaches are the ones that persist.
Using wireless connections, multiple devices can access cloud resources. The following are several common methods to synchronize sensor data to the cloud.
1. Wired Connection
This is the simplest method, developed in the 1970s and 1980s and the predecessor of wireless methods. Sensors include a microprocessor to process collected data, then upload the data over a wired network. The processor can also modify or update sensor functions. This method is limited because wired access is not available everywhere.
2. Cellular Network Connection
Cellular networks evolved after wired networks and became widely used. To connect to a base station, sensors typically connect via a phone or include an embedded baseband, which increases cost. Uplink transmitters require significant power, and data usage costs can be substantial.
3. Remote Wireless Network Connection
Unlicensed frequency bands were opened in the mid-20th century and gained value with mobile devices. Frequency bands such as 902-928 MHz and 2400-2483 MHz became popular for standards like IEEE 802.15.4. Mesh networks operating in these bands use many small, low-power wireless devices that aggregate sensor data from edge areas to collection points, which connect to the cloud to extend coverage.
4. Wireless Router Connection
The IEEE 802.11 Wi-Fi standard, first released in 1997, uses the 2400-2483 MHz and 5130-5835 MHz bands. Routers are widespread in homes, offices, and public places. Industrial or infrastructure-grade routers can connect to the cloud simply by attaching a wired uplink. With Wi-Fi on smartphones, sensors that can connect directly to routers have emerged, allowing sensors within router coverage to access the Internet without connecting to cellular base stations.
5. Connection via Mobile Devices
In many scenarios, sensors do not need to connect to a router; a mobile device can provide connectivity and direct interaction. Short-range wireless technologies such as Bluetooth operate in the 2400-2483 MHz band. Low-energy Bluetooth (BLE) offers reduced power consumption and is suitable for low-rate or low-duty-cycle simple sensors. BLE has accelerated the development of small sensors that previously required Wi-Fi or cellular connections, enabling direct interaction with mobile devices. Some routers have integrated Bluetooth, allowing BLE sensors to reach the cloud via the router without using a mobile device.
Conclusion
As the Internet of Things develops, sensor networks integrating cloud computing increase the level of data processing and the amount of information available. Further functional integration and fusion of capabilities is an expected trend and will support continued development of sensor technology.
