Bioscience goes wireless

Industrial Embedded Systems — November 13, 2008

1A biological specimen storage application shows how wireless sensor networks can be used to monitor large facilities with critical items that need constant care and produce copious amounts of data.

When Fisher BioServices, a leading biological specimen management company, wanted to reduce the use of manual logs in its facilities, it had two objectives in mind: increase productivity and continue to ensure specimen integrity for clients.

Based in Maryland with facilities around the world, Fisher BioServices provides pharmaceutical and biological support services for all sectors of the life sciences industry, including pharmaceutics, biotech, research, and government. Among its services are biorepositories (Figure 1) that offer secure, long-term biological sample storage. Until recently, these immense warehouses filled with rows of industrial refrigerators, freezers, and nitrogen-based cryogenic units relied on manual data collection to produce monthly product integrity reports, a tedious and time-consuming task.

Figure 1

Fisher BioServices provides clients with customized reporting. Stored samples are stringently monitored 24/7, and clients are sent monthly product integrity reports with detailed temperature logs. Historically, report data was collected manually from chart recorders connected to sensor probes in each refrigerator or freezer. Trained technicians periodically walked the warehouse floor checking the paper log on each chart recorder.

Although some hardwired alarm systems were in place, the old process did not offer centralized alarm monitoring capabilities. If a technician discovered that a refrigerator or freezer was out of specification, there was no other recourse but to destroy the samples and report the loss to the client.

Sensor nodes pass packets intelligently

In 2007, Fisher BioServices adopted the GE Sensing Kaye LabWatch wireless sensor network to automate data monitoring and collection at four of its biorepositories. LabWatch uses Dust Networks' embedded wireless network technology.

Figure 2

The mesh-based wireless sensor network comprises Kaye RF ValProbe sensor nodes and base stations (Figure 2). Each sensor node measures temperature via probes installed in a refrigerator or freezer and sends data through the wireless RF mesh to a base station connected to a physical Ethernet with servers running the Kaye LabWatch software. Up to 800 RF ValProbe sensor nodes and eight base stations are installed in each warehouse at the pilot facilities. The new system allows technicians to remotely view real-time data and compile historical sensor data for client reports, providing considerable savings in time and labor.

Ease of installation and scalability were two primary reasons why Fisher BioServices chose the RF ValProbe system. As Tim Wortley, global product manager for GE Sensing, remarked, "You turn it on, and it works".

These wireless ValProbe sensor nodes chirp at regular intervals, broadcasting a packet of collected data. This data is received by a nearby node, which in turn forwards the data to the next node. Data is thus transmitted from node to node until it reaches the base station. At the base station, the information is consolidated and passed through to the LabWatch system.

In a mesh network, nodes are installed in fairly close proximity to each other (typically less than 100 m apart), essentially blanketing an area. Each node has the intelligence to seek out and establish communication with other nodes in its network. It can then pass packets of information gathered by its own sensor or nearby devices.

Another important factor that Fisher BioServices considered when selecting the mesh network was its ability to reconfigure itself and transparently adapt to changing RF environments. "A hardwired monitoring solution would not have made sense for Fisher BioServices", Wortley said. "A successful repository operation must be able to move refrigerator and freezer units around in response to business loads and client requirements for segregated storage environments". Because the mesh network is self-organizing and self-healing, if one node goes down or is moved to another location, the system automatically finds the most optimal path to route sensor data to the base station.

Data reliability was also essential to Fisher BioServices because good manufacturing practices require the company to prove that specimen integrity is constantly maintained, thus necessitating accurate sensor data for client reports. Low power consumption was important, too, given that nodes are battery powered. The wireless mesh network does not require much power because each radio transmission travels only a short distance to reach nearby nodes and needs to communicate for just a brief period of time between measurements. In practice, this means that batteries can last for 5-10 years.

Self-configuration enables adjustments

The wireless network technology provided by the RF ValProbe nodes and base stations allows nodes to automatically discover their neighbors, self-organize into a mesh network, and begin reporting sensor data to the base station soon after they are installed. The network uses time-synchronized communication, which means that all nodes share a common sense of time and communicate in dynamic time slots. Routing algorithms dynamically balance the load across the network, further improving battery life.

Industrial warehouses are inherently unfriendly RF environments, and the Fisher BioServices sites are no exception. The concrete construction, massive refrigerators and freezers, constant reconfiguration, and movement of heavy equipment and people at these sites pose serious RF challenges. In addition, these facilities contend with the usual interference from other RF devices like Wi-Fi networks and Bluetooth cell phones. For a wireless mesh network to work reliably in such conditions, it must have three key characteristics:

  • Path diversity to provide multiple routes for communication to travel through the wireless mesh network to the base station and help navigate around physical obstructions
  • Frequency diversity to enable transmissions to hop across all available channels and navigate around potential RF interference
  • Time-synchronized transmissions with retries and acknowledgements to ensure that communications reach their destination and preserve precious battery life

Path diversity allows nodes to work their way around obstructions. A node is both a "parent" and a "child" or in radio terms, a receiver and a transmitter, and in a mesh network, every node has two parents. Having multiple parents provides path diversity. When a particular path is obstructed, a node can immediately respond to and navigate around it using an alternate path. This multipath configuration ensures that every node always has at least one backup (failover) communication path.

Intelligent self-configuration enables nodes to automatically adjust to a dynamic RF environment such as blocked signals and interference. The network is self-organizing and self-healing, meaning that when transmission is hindered, it rearranges itself and automatically finds a new path.

Self-configuration also supports a highly scalable network. A new node with the correct network ID can be added to the network simply by turning it on. And the network ID, which is included in all communications, allows multiple networks to operate in the same radio space without the risk of sharing data or misrouting information.

Channel hopping circumvents interference

Over time, this self-configuring ability drives overall and continuous network optimization. The network intelligently seeks the shortest, highest-quality transmission paths, reconfiguring as circumstances warrant.

In addition to multiple path options, nodes have channel-hopping options to work around intermittent RF interferences. Each child-to-parent connection is called a path. Depending on the radio standard used, each path has a number of channels. For example, using the radio standard IEEE 802.15.4, each path has 16 channels available for communication (think 16 radio stations on a radio dial). Channel hopping means the nodes use all 16 channels available on the IEEE 802.15.4 wireless radio to avoid RF interference. The node uses a different channel for each transmission, creating a mesh network that is exceptionally reliable and resilient in harsh and variable RF environments. Hopping across multiple channels is a proven method for sidestepping interference and overcoming RF challenges with agility rather than brute power.

This frequency diversity enables transmissions to hop across all available channels and thus navigate RF interference. Packets are re-sent until they are successfully transmitted. The algorithm that sends packets from the child to the parent, trying a different channel if it does not get through or sending to a different parent if it still does not get through, significantly improves the system's overall reliability.

With nearly a thousand sensor nodes already installed at its Maryland facilities, Fisher BioServices intends to deploy globally in the next three years, with plans to install wireless sensor networks at facilities in South Africa, the United Kingdom, and Singapore.

Steve Toteda is VP of marketing for Dust Networks, based in Hayward, California. Steve has more than 18 years of marketing and technical experience, having worked at companies including Cisco Systems, Komodo Technology, LSI Logic, Cirrus Logic, and Harmonic Lightwaves. In addition to his role at Dust Networks, Steve serves as a board member of the Wireless Industrial Networking Alliance. Steve holds BS and MS degrees in Engineering from Lehigh University and an MBA from Columbia Business School.

Dust Networks