In 2015, reports began to appear of a “Passive WiFi” technology emerging from the University of Washington under the direction of Dr. Shyam Gollakota. Passive WiFi used backscatter techniques to enable ultra-low power Internet of Things endpoints that could transmit sensor data using ambient HDTV broadcast signals. The key innovation was the ability to passively reflected the signals to appear to standard WiFi receivers like dynamic WiFi packets.
When ambient RF energy proved to be too unreliable, the researchers returned in 2017 to announce a new version with a similar long-range backscatter technology that instead reflected dedicated carrier signals built on top of the LoRa wireless protocol. The university also announced that it had launched a spin-off called Jeeva Wireless to bring the technology to market.
Jeeva Wireless has now unveiled a variation of the non-ambient technology in the form of a low-cost IC called Parsair™ that implements the backscatter modem on an IoT endpoint. The Seattle-based company recently launched an evaluation kit that packages Parsair™ in an endpoint device. The kit includes a pair of “companions” based on commodity RF technology to act as the RF source and backscatter receiver.
“Jeeva is the only currently available wireless technology that can meet the power, cost, size, and complexity requirements that are necessary to scale the growing demand for connected devices,” says Scott Bright, CEO at Jeeva Wireless, a member of MIT’s STEX25 accelerator for innovative startups. “We use exponentially less power than any existing wireless technology, for only a small fraction of the usual component cost.”
Jeeva is the only currently available wireless technology that can meet the power, cost, size, and complexity requirements that are necessary to scale the growing demand for connected devices.
Jeeva bridges the gulf between short-range, low-bandwidth RFID and more robust and longer-range wireless protocols that use symmetrically emitted RF such as Bluetooth, ZigBee, LoRa, and WiFi. Passive RFID, which uses a much simpler form of backscatter, has the advantage of a very low-cost endpoint. However, “the reader antennas can be expensive and require close proximity to the RFID tags,” says Bright. Wireless technologies, meanwhile, are far more expensive and power-hungry than RFID. Commodity RF endpoints cost about $5 to $6 and consume 5-10 mW.
“Jeeva brings together the best of both technologies,” says Bright. “For the same cost, size, and power as a passive RFID tag, we can transmit from hundreds of kilobits per second up to a megabit per second of data across a range of up to 100 meters while using less than a quarter of a milliwatt.”
Jeeva Wireless possesses 24 issued and pending patents and owns an exclusive, unlimited, global license to related wireless technology coming out of the University of Washington R&D lab. In recent years, Jeeva has been fine-tuning the platform-agnostic technology and sorting out interference issues, primarily from other wireless radios wiping out the weak backscatter signal.
The interference problems have been solved with technologies like frequency hopping, adaptive power control, and CSMA. The company has now ramped up its pilot projects for applications including warehouses, supply chain, cold chain tacking, logistics, and industrial automation. Medical and consumer packaged-goods partnerships are also in the works.
“At long last we are on the cusp of commercialization,” says Bright. “We plan to scale up to volume later this year.”
The Jeeva companion radio issues a single tone of RF energy in the 840-960 MHz range that is utilized by multiple Jeeva endpoints. The endpoints dynamically reflect this signal using a variation on chirp spread spectrum (CSS) and/or phase-shift keying (PSK) to encode sensor data for delivery to one or more “companion” receiver radios. The signal can be encoded with a variety of standard sub-GHz protocols, including LoRa and ZigBee.
The evaluation kit uses a half-duplex topology, in which a companion unit “illuminates the area with the carrier signal” and an identical companion receives the reflected signal from the endpoint, says Bright. “The sender and receiver companions can dynamically trade places if the endpoint is moving around.” The Jeeva technology also supports a full-duplex topology, in which the source and receiver functions are combined in a single, but slightly more expensive, companion.
In some cases, the Jeeva backscatter system can be implemented on top of an existing network, with radios sending and receiving Jeeva signals when not otherwise in use. “In theory, any commercial radio that provides low-level access to the embedded operating code can be made to implement a backscatter network,” says Bright. “In practice, access to the low-level code can become a constraint. We use our own commodity hardware because it’s easier to control.”
Jeeva has a data rate of up to 1Mbps and for most applications, a practical range of about 100 meters. “If you need to send smaller amounts of data, we can support a range between companions of up to 200 meters,” says Bright.
Almost all the power required for transmitting the reflected signal comes from the companion’s RF carrier signal, enabling endpoint consumption of 0.2 milliwatts (mW). By comparison, Bluetooth Low Energy (BLE) uses 5-10 mW and ZigBee eats up 10-50 mW.
“In a traditional emissive radio system, there is a tradeoff between bandwidth and power,” says Bright. “What’s novel about a backscatter network is that power is fixed – it takes the same amount of power to send a megabit of data as it does to send a kilobit, and whether the range is long or short.”
Instead, the tradeoff becomes range vs. throughput. “At room scale, our networks can send at up to 1Mbps, but to cover hundreds of meters, the data rate goes down to the 10s to 100s of Kbps.”
The endpoints supplied in Jeeva’s eval kits package the Parsair™ chip with an antenna and a low-cost microcontroller, which passes sensor inputs to the modem via a serial interface. There is also a coin-cell battery that, depending on the use-case, can last for several years. Customers with remote or inaccessible sensor endpoints can instead power up with energy harvesting technologies such as thermal, solar photovoltaic, kinetic, and RF harvesting. “It is far more challenging for an emissive radio endpoint to use these sources,” says Bright.
In high-volume quantities, a complete Jeeva endpoint costs about half of what a traditional endpoint might cost,” says Bright.
Another Jeeva advantage is that it supports dense endpoint deployment. A single companion can theoretically communicate with up to 65,000 endpoints.
“Practically, that upper limit is limited by the frequency of transmission and throughput requirements,” explains Bright. “For a pallet tracking or cold-chain tracking network that does not require real-time information, we can build networks with thousands of devices per companion. If you need real-time input, perhaps for industrial process control or predictive maintenance, you could have a range up to 10s of meters, in which case you would probably need to deploy companions on a grid or mesh system.”
One advantage over RFID is the ability to transmit from deep within a pallet. “The only way to do that with RFID is to bring the reader in very close proximity, and it would be very challenging,” says Bright. “Our sub-GHz technology offers deep penetration and ease of operation in RF environments with line-of-sight occlusion challenges.”
The companion typically sends a uniform carrier signal rather than an encoded signal. However, a downlink capability is available that can remotely wake endpoints from a low power state to send data. “On a traditional radio network in which data is sent intermittently, most of the power is used to stay awake to listen for a command signal,” says Bright. “By remotely waking an endpoint from a single-digit microwatt state, we can have very dense deployments with greater energy efficiency.”
Jeeva also has a provision for manufacturing modems with a unique 8-byte address for ID purposes. Combined with the bidirectional capability, “the unique IDs enable configuration control or the ability to command specific sensors to go into different modes,” says Bright.
Jeeva’s pilot partners in supply chain, logistics, and automation are primarily using the technology to “harvest information globally about how their products are being used,” says Bright. “We aggregate the information into a cloud datastore and provide a mobile app and a dashboard interface so our customers can access it and bring it into their own supply and distribution chains to enable efficiency gains.”
We aggregate the information into a cloud datastore and provide a mobile app and a dashboard interface so our customers can access it and bring it into their own supply and distribution chains to enable efficiency gains.
The company is currently investigating medical applications with the help of NSF funding. “We are looking into embedding our modem in disposable packaging for consumable medical supplies,” says Bright. “Our technology could track when a package has been removed from a shelf and opened and then inform the distribution chain to enable automated re-fulfillment.”
Jeeva is working with a consumer goods company to demonstrate a similar application in packaged consumables. “Imagine your laundry detergent bottle with a built-in sensor that can communicate over WiFi to tell you when it’s time to re-order,” says Bright.
Another customer is looking to embed the modems in disposable insulin injector pens. “We showed how the pen can communicate with a cellphone over Bluetooth and relay usage and dosage data so healthcare providers have real-time insights into product usage.”
Jeeva recently engaged in a research project with NASA to see if its backscatter technology could achieve the same timing specifications possible with a wired network. “On a spacecraft or aircraft, sensor data needs to be correlated in time, which is easier on a wired network since all the data is traveling at basically the same speed,” says Bright. “With a packet-switched wireless network, there’s an inherent indeterminacy as to when the data was received. You need to know when it was generated to correlate it. We proved we could correlate that to the same precision possible with a wired network.”
The research could enable NASA to replace heavy wiring with wireless networks in future spacecraft. It could also help Jeeva further reduce the cost of its endpoints. “We may be able to avoid the need for a crystal oscillator to establish the time source,” says Bright. “We think we can remotely sync using our network protocol alone.”
Although Jeeva is optimized for sensors with low-bandwidth data, there is the potential to expand upward. “Voice audio is within the capability of our current network, and we expect to scale up to low-bandwidth video in our next generation,” says Bright.