One of the most popular entry points for retailers as far as two-way radio is concerned is PMR446, given that it allows them to benefit from instant one-to-many group calls while being simple, low-cost and licence-free.
“PMR446 is perfect for a lot of retail stores because they don’t need to buy big infrastructure, they don’t need a repeating system, just small, cheap 446s,” says Lawrence Deacon, product support engineer at South Midlands Communications (SMC). “A lot of people are opting for PTT over Cellular to get away from frequency spectrum overload. PMR446 is another good way of doing that.”
With the growing ease at which two-way radios can be connected to IT systems, there is more that can be done with them, and SMC has been working on new features, using a Motorola XT420 PMR446 radio connected to its SMC Gateway, which it will be launching soon.
The company can play audio messages over the PMR446 handset to others in the network, and the audio can be generated in multiple ways, commonly by SMC’s text-to-speech engine and recorded audio stored on the Gateway. To get around the limited capacity given lack of network infrastructure and the absence of dedicated spectrum, SMC uses PTT detection and stores the messages in a buffer – “as soon as the PTT is released from someone else, we can fire the messages [off], but we can queue them indefinitely”, Deacon adds.
Some of the uses include call buttons in fitting rooms that when pressed result in an audio message, based on the button’s location, to alert a member of staff that a customer needs assistance. Alarms can be relayed via audio, either text to speech or pre-recorded, to allow smaller licence-free radios to benefit from alarm notifications. Deacon says it can also be used to alert staff to important emails – those that contain a keyword – by reading out the subject line over the radio.
Another use-case would be to use the system in combination with motion-detection equipment to allow salespeople on a large shop floor to come over when a customer is interested in a particular high-value item, rather than having to stand near it waiting for customers.
“You don’t want someone stood there all day long, but when someone starts looking at a washing machine (for example, starts opening its doors), you might want to send a salesman over there. There’s no reason why a motion sensor couldn’t be set up so that when a customer moves into the area, someone can have a look and then move there to upsell,” Deacon says. And security guards can be alerted if the product is often targeted by shoplifters.
Lighting up the light-fingered
While we are on the subject of preventing shoplifting, I recently revisited Facewatch’s offices to hear how its facial-recognition work has progressed since I interviewed Simon Gordon roughly a year ago (see January 2017’s Big Interview). Its new CEO Nick Fisher explains that it has changed its focus to bringing down the cost of facial-recognition technology to the point where such systems can be used as an alternative or complement to traditional CCTV cameras in retail stores.
“Our aim was to significantly reduce the price to the customer to make Facewatch easily affordable and through extensive development over the last 12 months we’re now at one third of the price we were 12 months ago,” he says. Fisher adds that a single edge server will be able to handle a convenience store with a footfall of 1,200-1,800 people a day. He expects that most businesses “will buy the cameras, server and installation as they can depreciate it as an asset over at least five years” and rent the licence for the use of the software.
The technology that Facewatch uses can work out a person’s gender, ethnicity and age (give or take five years). While this could be used to generate a great deal of insight for retailers, Fisher explains that it isn’t pursuing this approach, choosing instead to store only faces on its systems, which use “bank-level security” and encryption on all data at rest. Facewatch’s proposition is for a camera to be mounted near to a premises’ entrance. Facial-recognition algorithms that are designed to work in low-light conditions and are run on a local edge server detect whether face(s) in the camera’s field of vision are on a watchlist, and only those that are are then sent to Facewatch’s cloud service, which is hosted in the UK. A yes/no alert is then sent to staff through whatever method they request. The data that needs to be processed hinges on footfall, requiring the amount of computing power to be scaled appropriately.
Facewatch’s facial-recognition system in action
Fisher explains that the idea here is to replicate Facewatch’s pre-existing proposition (the sharing of watchlists between stores in the same area) in a way that eliminates the need for staff to remember the faces of known shoplifters. He adds that this is much needed due to rapid staff turnover in retail.
Similarly, Fisher highlights the fact that the majority of petty criminals are inherently lazy, keep to their local area and share information with each other. He expects that the first time they get picked up by a store using facial recognition, they will recount their experience to their peers, increasing its value as a deterrent.
Fisher is confident that such a system can reduce in-store crime by a conservative 35 per cent in the first year, given the 40 per cent reduction at a big grocery chain that has been using Facewatch’s static watchlist for three years. He adds that while the company’s facial-recognition solution is due to be formally launched in January, it is currently available and Facewatch is about to run a couple of trials of the technology – one in a national food chain, another with a petrol retailer/convenience store, and one in a “prestigious hotel in central London”.
The latter is considering other uses for the technology. “They have to speak to their lawyers about using it as a white list,” Fisher says, “to identify celebrities and public figures who might be of value.”
Micro-proximity, big data
Turning to cellular, retailers can glean a great deal of insight from the location data that mobile operators gather from their subscribers’ devices. However, the accuracy of location data from macro-cells is an issue, as is the problem of in-building coverage.
This is where small cells come in. Back in November, ip.access announced that it had signed a partnership agreement with O2 to deliver micro-proximity customer analytics through the ip.access PRESENCE solution. Nick Johnson, ip.access’s CTO, explains that his company’s PRESENCE sensors are low-power versions of one of its small cells. “It only reaches out a few metres, maybe 20 or 30 metres maximum. You can tune that power to get the range you desire, whether you want to know whether someone is within five metres of a display or whether someone is somewhere within the store.
“The information can be transmitted back to the collector in real time or it can be batched up and sent once or twice a day. Physically it looks a bit different from the kind of thing we normally provide; it’s just a box with a power lead, the customer can just plug it where they’re able to and where it makes most sense for the thing they’re trying to find out.”
He adds that by putting one of these sensors in a display would allow retailers to know “how long people linger in front of it and then what they do next – do they go to the counter and buy one or do they just wander around the shop? You can learn little bits of insight into how they behave, how effective the display is. Take an advertising display, for instance. Does someone actually respond to it within days or weeks of being exposed to it? How much impact did that advertising have?”
Johnson says that when working with operators, ip.access’s sensors are a data source that feeds into the operator’s “enrichment stack”, which can then be enhanced by the operator’s own data on subscribers, “depending on the permissions and opt-in/opt-out [settings]”.
One application that ip.access has provided for retailers involves programming a particular small cell to send an alert when a certain individual arrives. “With a lot of our operator customers, the first place where this technology is applied is in their own retail stores, so you can program an individual cell to [tell you] when a certain individual arrives; the store manager will get a notification,” Johnson says.
“It gives you a bit of a leg-up when it comes to customer service; you can apply that same application to individuals, as someone approaches the counter – you’ll have a little bit of advance information about who they are, just making the customer experience better and smoother and more personal, again without infringing anyone’s privacy.”
Speaking of privacy, Johnson says: “Although we collect an IMSI [international mobile subscriber identity] which is identifiable by an operator, when we collect it and transfer it to them, it’s done in an encrypted and secure way so we don’t know who the person is – and in the form that it’s transported, any man in the middle who might intercept that data won’t know who it is either.
“Then the operator applies their own opt-in and opt-out rules to that little bit of data. The system as a whole then applies to GDPR and those privacy permission requirements. If a subscriber has opted into a loyalty scheme then the operator may be free to pass the movements of that particular individual onto their customers in an identifiable way, according to their preferences.”
He adds that in the absence of such permissions, the data is used in an aggregated way.
Johnson says that an operator might be able to share a great deal of information about a subscriber with a retailer, such as whether they visited Amazon’s website immediately after going to one of the retailer’s stores, and this sort of data could also be available on an aggregated/anonymised basis.
We have seen that even the humble PMR446 radio can now be used to improve situational awareness on the shop floor and automatically transmit audio messages and alerts, while facial recognition can help tackle shoplifting, not to mention the huge amount of insight that can be obtained from our mobile phones to help retailers fine-tune their stores and marketing efforts. While, as PWC has recently highlighted, 2017 has been a challenging year for retailers, perhaps they can take some comfort in this wave of innovation.
Getting out and about
While the focus of this piece is on retail, ip.access’s Nick Johnson says that transport is another sector that can benefit from the big data gathered by mobile networks. He adds that with this application there is no need to use any of the richer data that can be obtained from operators’ contractual relationship with their transcribers.
“Because the identifier we collect, the IMSI, is permanent, it doesn’t change from sighting to sighting, compared with Bluetooth or Wi-Fi type techniques; you can get very useful information about traffic flows, people flows through transport networks, without infringing anyone’s privacy as the IMSI isn’t used for any marketing purpose. But you do know that [someone] arrived at a train station, took a journey, took a bus and was held up, spent a long time waiting in a queue at a station and so on. Aggregating that kind of data gives you a lot of useful stuff that can help improve the performance of your transport network. You’re not interested in figuring out where individuals [go] but you get a lot of powerful information about how to minimise journey and wait times in general.”
