Using cb library to connect simultaneously multiple TI SensorTag via Bluetooth Low Energy

ProgrammingGo

Thanks a will try it out :).
I have a question regarding the example of the SensorTag which is given in the documentation of cb:

   def did_discover_peripheral(self, p):
        print('+++ Discovered peripheral: %s (%s)' % (p.name, p.uuid))
        if p.name and 'Sensor Tag' in p.name and not self.peripheral:
            # Keep a reference to the peripheral, so it doesn't get garbage-collected:
            self.peripheral = p
            cb.connect_peripheral(self.peripheral)

This is the function to detect peripherals which are advertising. In the if statement there is something which is a bit unclear. Does it mean that if p.name is available and SensorTag is in P.name is clear, but why not self.peripheral?

omz

@ProgrammingGo

In the if statement there is something which is a bit unclear. Does it mean that if p.name is available and SensorTag is in P.name is clear, but why not self.peripheral?

The example connects to just one SensorTag, and that SensorTag is stored in the self.peripheral attribute, so the if not self.peripheral condition means "ignore this if we already found it". As you want to connect to multiple SensorTags, you'd have to handle this differently, of course.

@ccc's code from earlier should be a good starting point. You might just need to use UUIDs instead of names, if all your SensorTags have the same name (not sure if they do generally).

ProgrammingGo

Hi,
okay thanks for the explanation. I will try it out. that means the dict {} stores the connected peripherals so that the central does not try to connect to them again. I saw that the dict {} in the code is empty, should this stay like this or can I define the names of the nodes to which I want to have the connection?
Regarding the name I can change the name of the SensorTag so that I could distinguish them

ccc

When you start the dict is empty. Each time that did_discover_peripheral() is run, it will add a new entry in the dict if it is not already there.

ccc

@ProgrammingGo Please run the following with multiple SensorTags turned on and post the output here:

class MultipleDeviceManager(object):
    def __init__(self):
        self.by_name = {}  # dict of peripheral names and peripheral objects
        self.by_uuid = {}  # dict of peripheral uuids and peripheral objects

    def did_discover_peripheral(self, p):
        if p.name not in self.by_name:
            self.by_name[p.name] = p
            print(f'Connecting to a new device: {p.name}...')
        if p.uuid not in self.by_uuid :
            self.by_uuid[p.uuid] = p
            print(f'Connecting to a new device: {p.uuid}...')
            cb.connect_peripheral(p)
        print(f'{len(self.by_name)} unique names and {len(self.by_uuid)} unique uuids')

    def connect_to_device(self, device_name):
        """May raise a KeyError"""
        cb.connect_peripheral(self.by_name[device_name])

ProgrammingGo

@ccc I tried it out with two tags ( using the same name for both tags) and I have the following output:

Output:

Connecting to new device: CC2650 SensorTag....
Connecting to new device: <UUID> (too long but was displayed)

1 unique names and 1 unique uuids

Connecting to new device: <UUID> ---> only uuid was displayed because second SensorTag has same name

1 unique names and 2 unique uuids.

So logically I could use the uuid to have multiple sensor connected. I see the problem for the callback functions where I try to read out the sensor values or update it. For services it works, but to read out the same characteristics from both tags it is a problem. I would need to have a reference of the peripheral in the callback function did_discover_services. I wanted to modify the function, is there another way as workaround?

ccc

This is exactly what I anticipated that you would find... All these devices share a common name but each one has a unique UUID.

import cb


class MultipleCC2650Manager(object):
    name = 'CC2650 SensorTag'  # we only care about devices with this name

    def __init__(self):
        self.devices = {}  # dict of peripheral uuids and peripheral objects

    def did_discover_peripheral(self, p):
        if p.name != self.name:
            return  # ignore devices do not have the name 'CC2650 SensorTag'
        if p.uuid in self.devices:
            return  # ignore devices that we have already discovered/registered
        self.devices[p.uuid] = p
        print(f'Connecting to a new device: {p.uuid}...')
        cb.connect_peripheral(p)
        print(f'{len(self.devices)} unique {self.name} devices')

    def connect_to_device(self, device_uuid):
        cb.connect_peripheral(self.devices[device_uuid])

JonB

One thing you have to watch out for is that if you are trying to do notify in characteristics, the callbacks dont tell you whch peripheral it came from. That is a major shortcoming in my mind of the cb implementation. You might be better off going to an objc_util based solution

ProgrammingGo

@ccc Yeah thanks a lot. I understood it very well with your example. Thank you so much. So I need to work with the uuid of the devices. A question regarding the example given : you call cb.connect_peripheral(p) in did_discover_peripheral and then in def connect_to_device again,why?

delegate = MyCentralManagerDelegate()
print('Scanning for peripherals...')
cb.set_central_delegate(delegate)
cb.scan_for_peripherals()

# Keep the connection alive until the 'Stop' button is pressed:
try:
    while True: pass
except KeyboardInterrupt:
    # Disconnect everything:
    cb.reset()

This part is still needed right?

ProgrammingGo

@JonB Ok thank you for your hint

ccc

did_discover_peripheral() will always leave you connected to the last device that it discovered. So when you want to chose which device that you want to connect to, you call connect_to_device().

ProgrammingGo

@ccc Hi sorry for the question I am a bit confused. It is not clear to me how do you mean that :(. Can you explain it may be again? My target is to be connected to 3 SensorTags at the same time to receive data.

ccc

Let's say that you have 3 CC2650 SensorTags and their UUIDs are 0, 1, 2... Getting data would be:

data = []  # empty list
for device_uuid in (0, 1, 2):
    cc2650_manager.connect_to_device(device_uuid)
    data.append(get_data())
print(data)

You can only talk to one device at a time but you can loop thru all the devices and call connect_to_device() and then get the data from that device.

ccc

Do you have GitHub, GitLab, or BitBucket? It would probably be easier to collaborate on a repo.

ProgrammingGo

@ccc Ok that means I am using a loop to read node by node. But for me, it is not so clear how to use connect_to_periphera to connect to the next. - How can I call it? For my understanding normally at first, I would call connect_to_peripheral and then did_discover_peripheral.
I gave a look in the example given for SensorTag and calling the callback functions is a bit different ---> it happens inside the cb lib.

No Iam not using GitHub or the other tools.

JonB

You can connect to each device once, as long as you hang into references to the peripheral object and Characteristic.

The trick is then keeping track of the reads, since did_update_value does not include the peripheral, so you cannot tell which is which.

so pseudo code would be:

discover devices
connect to devices, storing peripheral in a list or dict
discover service/characteristic, and store the characteristic
once you have a characteristic, i would have a loop that pools:

for p in peripherals:
   current_p=p (store as a global)
   p.read_characteristic_value(C)
   (sleep for a while p, or use threading Event etc to await did_update_value)

in did_update_value, check current_p to get uuid, then store the read value somewhere, print it out, whatever, and set the Event flag to awake the main loop

ProgrammingGo

@JonB Hi thanks for the nice description: I tried to work with the structure of the given example for the sensortag Example and try it out to adapt it. But Iam getting stuck. Iam not able to read out the data from all 3 sensortags. When I try to register for characteristics nothing worked. I tried several things out but without success.

ProgrammingGo

@JonB said:

You can connect to each device once, as long as you hang into references to the peripheral object and Characteristic.

The trick is then keeping track of the reads, since did_update_value does not include the peripheral, so you cannot tell which is which.

so pseudo code would be:

discover devices

connect to devices, storing peripheral in a list or dict

discover service/characteristic, and store the characteristic

once you have a characteristic, i would have a loop that pools:
for p in peripherals:
   current_p=p (store as a global)
   p.read_characteristic_value(C)
   (sleep for a while p, or use threading Event etc to await did_update_value)
in did_update_value, check current_p to get uuid, then store the read value somewhere, print it out, whatever, and set the Event flag to awake the main loop

The steps from the pseudo code are understandable but do you mean storing the discovered peripherals in a list and use a for loop to connect one by one and read the data or do you mean connect to all and use a for loop to read out the data of one device per time`?

JonB

https://gist.github.com/f734504daa44d0cb0621ba47c2caabb0
Here is a working example which:
tries to connects to every peripheral discovered
discovers all characteristics of all services found
polls all characteristics every 6 seconds, and displays the results

type stop() to stop the polling. then you can access peripherals via
m[uuid] or m[index]. the cb.peripheral is in the peripheral field.

i.e
m[0].peripheral.read_characteristic_value(m[0]['180A']['2A24'].characteristic)

print(m[0]['180A']['2A24'].characteristic)

or, use m.update_all() to poll all characteristics, and print(m) to show them all.

JonB

what i have found is that you can indeed connect to multiple peripherals. but sometimes cb decides to disconnect. or sometimes decides not to discover a peripheral at all. cb.reset() can help, but things just seem really flaky, and a pythonista reset might be better. for instance, it seems like if a connection fails enough times, it is hidden from the app and wont be discovered again.