BLE channels

The BLE system operates in the 2.4 GHz ISM band at 2400 - 2483.5 MHz. It uses forty RF channels (each channel is 2 MHz wide). The figure below shows the mapping between the frequencies and BLE channels. Each of these RF channels is allocated a unique channel index (labelled as "BLE Channel" in the figure).

BLE classifies these forty RF channels into three advertising channels (channel indices: 37, 38, 39) and thirty-seven data channels (channel indices: 0 to 36). Note that the advertising channels are spread across the 2.4 GHz spectrum. The purpose of this wide spacing is to avoid interference from other devices operating in the same spectrum, such as WLAN. Advertising channels are mainly used for transmitting advertising packets, scan request/response packets and connection indication packets. Data channels are mainly used for exchanging data packets.

Channel Hopping

Channel hopping is used in Bluetooth to reduce interference and improve throughput. The Bluetooth standard defines rules for switching between channels and algorithms used when performing channel hopping.

Use of the unlicensed 2.4GHz ISM band by several wireless technologies causes increased interference and results in retransmissions to correct errors in received packets. Since BLE is a low energy oriented protocol, it is more susceptible to interference. BLE uses channel hopping to combat the impact of interference. When one channel is completely blocked due to interference, devices can still continue to communicate with each other on other channels.

In classic Bluetooth, channel hopping is restricted to 1600 frequency hops/sec. For BLE, the channel hopping specification has been revised. Different rules apply for advertising and connected devices, and two channel selection algorithms are defined.

An advertising device transmits advertising packets on the three advertising channels in a cyclic manner (starting from channel index 37). The same procedure is used by the scanning/initiating device, listening on the three advertising channels in a cyclic manner.

A connected device changes to a new data channel for every connection event. A connection event is a sequence of data packet exchanges between two connected devices. The connection events occur periodically with an interval called connection interval. All the packets within a connection event are transmitted on the same data channel. A new connection event uses a new data channel.

Two alternative channel selection algorithms are specified by the Bluetooth core specification (see Section 4.5.8, Part-B, Vol-6 of [ 1 ]) can be used to select data channels for each connection event:

The two channel selection algorithms avoid channels that are prone to transmission errors. A channel map is exchanged between the master and slave devices. This map indicates the good and bad data channels. The classification of good and bad data channels is implementation dependent and can be done based on various parameters like SNR (Signal-To-Noise Ratio), PER (Packet Error Rate), etc. Only the good data channels are used for communication between devices. The channel map will be updated by the master device if it recognizes any bad data channels. The two channel selection algorithms use the channel map to determine whether the selected data channel is good to use. If the selected data channel turns out to be bad, a new data channel is selected using channel remapping procedure (see Section 4.5.8, Part-B, Vol-6 of [ 1 ]), which remaps the bad data channel to one of the good data channels. Each algorithm has a remapping procedure of its own.

Check for Support Package Installation

% Check if the 'Communications Toolbox Library for the Bluetooth Protocol'
% support package is installed or not.
commSupportPackageCheck('BLUETOOTH');

Simulating Algorithm #1

You can use bleChannelSelection System object to select a new channel index. This System object configures the fields required for selecting a channel index.

Create a System object for 'Algorithm #1'

To select a channel index, create a bleChannelSelection System object with Algorithm set to 1.

csa = bleChannelSelection('Algorithm', 1);

Configure the fields.

csa.HopIncrement = 8;
csa.UsedChannels = [0, 5, 13, 9, 24, 36]

csa = 

  bleChannelSelection with properties:

       Algorithm: 1
    HopIncrement: 8
    UsedChannels: [0 5 9 13 24 36]
    ChannelIndex: 0
    EventCounter: 0

Select a channel index for next hop

Call the object csa as a function to determine the next channel hop and to select a new channel for each new connection event.

nextChannel = csa();
fprintf('Selected channel for connection event %d using ''Algorithm #1'' is: %d\n', csa.EventCounter, nextChannel);

Selected channel for connection event 1 using 'Algorithm #1' is: 9

Simulating Algorithm #2

You can use bleChannelSelection System object to select a new channel index. This System object configures the fields required for selecting a channel index.

Create a System object for 'Algorithm #2'

To select a channel index, Create a bleChannelSelection System object with Algorithm set to 2.

csa = bleChannelSelection('Algorithm', 2);

Configure the fields.

csa.AccessAddress = 'E89BED68';
csa.UsedChannels = [9, 10, 21, 22, 23, 33, 34, 35, 36]

csa = 

  bleChannelSelection with properties:

        Algorithm: 2
    AccessAddress: 'E89BED68'
     UsedChannels: [9 10 21 22 23 33 34 35 36]
     ChannelIndex: 0
     EventCounter: 0

Select a channel index for next hop

Call the object csa as a function to determine the next channel hop and to select a new channel for each new connection event.

nextChannel = csa();
fprintf('Selected channel for connection event %d using ''Algorithm #2'' is: %d\n', csa.EventCounter, nextChannel);

Selected channel for connection event 1 using 'Algorithm #2' is: 21

GUI for analyzing Channel Selection Algorithms

The function helperBLEChannelHopSelectionUI provides a graphical user interface to generate desired number of channel hops for analyzing the algorithm. Both channel selection algorithms can be analyzed using this GUI. It can be used to plot the channel hopping pattern of an algorithm and also plots the corresponding histogram.

helperBLEChannelHopSelectionUI()

Algorithm verification with sample data

Sample data is provided to verify Algorithm #2 (see Section 3, Vol 6, Part C in [ 1 ]). However, there is no sample data available for verifying Algorithm #1.

Sample data 1 (thirty-seven good data channels)

When the above inputs are used, the Algorithm #2 is expected to select the following channels according to Section 3.1, Part-B, Vol-6 of [ 1 ]

The following code selects three channels for the first three connection events.

% Create a System object for 'Algorithm #2'
csa = bleChannelSelection('Algorithm', 2);

Configure the fields with sample data #1.

% Connection access address
csa.AccessAddress = '8E89BED6';
% Use 37 good data channels as used channels according to the sample data
csa.UsedChannels = (0:36);

Select channel indices for first 3 connection events. Verify the generated outputs with the table mentioned above.

numConnectionEvents = 3;
for i = 1:numConnectionEvents
    channel = csa();
    fprintf('Event Counter: %d, selected Channel: %d\n', csa.EventCounter, channel);
end

Event Counter: 1, selected Channel: 20
Event Counter: 2, selected Channel: 6
Event Counter: 3, selected Channel: 21

Sample data 2 (nine good data channels)

When the above inputs are used, the Algorithm #2 is expected to select the following channels according to Section 3.2, Part-B, Vol-6 of [ 1 ]. Since the channel map contains bad channels, the channel remapping procedure used in the algorithm is also verified.

The following code selects eight channels for the first eight connection events.

% Create a System object for 'Algorithm #2'
csa = bleChannelSelection('Algorithm', 2);

Configure the fields with sample data #2.

% Connection access address
csa.AccessAddress = '8E89BED6';
% Use 9 good data channels as used channels according to the sample data
csa.UsedChannels = [9, 10, 21, 22, 23, 33, 34, 35, 36];

Select channel indices for first 8 connection events. Verify the generated outputs with the table mentioned above.

numConnectionEvents = 8;
for i = 1:numConnectionEvents
    channel = csa();
    fprintf('Event Counter: %d, selected Channel: %d\n', csa.EventCounter, channel);
end

Event Counter: 1, selected Channel: 9
Event Counter: 2, selected Channel: 33
Event Counter: 3, selected Channel: 21
Event Counter: 4, selected Channel: 34
Event Counter: 5, selected Channel: 36
Event Counter: 6, selected Channel: 23
Event Counter: 7, selected Channel: 9
Event Counter: 8, selected Channel: 34

Plot and analyze the hopping pattern - Algorithm #1 and Algorithm #2

The following code selects channel indices for the first hundred connection events using 'Algorithm #1'. The selected channels are plotted and compared with those of 'Algorithm #2'.

% Channel selection algorithm System object for 'Algorithm #1'
csa = bleChannelSelection;
% For 100 connection events
numConnectionEvents = 100;
hopSequence = zeros(1, numConnectionEvents);
% Generate channel hop sequence for 100 connection events
for i = 1:numConnectionEvents
    hopSequence(i) = csa();
end

The helperBLEPlotChannelHopSequence function plots the hopping pattern and also outputs a histogram of the selected channels.

helperBLEPlotChannelHopSequence(csa, hopSequence);

The following code generates channel indices for the first hundred connection events using 'Algorithm #2'. The selected channels are plotted and compared with those of 'Algorithm #1'.

% Channel selection algorithm System object for 'Algorithm #2'
csa = bleChannelSelection('Algorithm', 2);
% For 100 connection events
numConnectionEvents = 100;
hopSequence = zeros(1, numConnectionEvents);
% Generate channel hop sequence for 100 connection events
for i = 1:numConnectionEvents
    hopSequence(i) = csa();
end

The helperBLEPlotChannelHopSequence function plots the hopping pattern and also outputs a histogram of the selected channels.

helperBLEPlotChannelHopSequence(csa, hopSequence);

Algorithm #1 vs Algorithm #2

The above plots show the difference between the two algorithms.

Conclusion

This example demonstrated the behavior of the channel selection algorithms specified in the Bluetooth core specification [ 1 ].

Appendix

This example uses the following helper functions

Selected Bibliography