This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Is TI MAC (802.15.4) suitable for my application?


Hi, all experts:
I would like to use TI MAC for my application, but I wonder if it could meet my requirements:
1. There are about 2000 ED (End Device, RFD in the spec) in one PAN. There might be more than one PAN in the system, and different PAN might need to use different channels to avoid interference to each other. Let’s call PAN coordinator as AP in the following.


2. When AP wants to send some data to an ED (the data size is about several k bytes, let’s suppose it is 5K), it is expected to complete the transit within 1 or 2 minutes. So I think it has to be a beacon-enabled PAN, otherwise it would not be possible for so many ED to query data from AP within so short time. The length of the superframe should be about 1 or 2 minutes. The max length of superframe that 802.15.4 supports is about 4 minutes (when Superframe Order is 14).

3. When AP needs to send data to many ED (Let’s say all 2000 ED), it should be completed within 30 minutes. So the average effective data rate is (2000 * 5k bytes)/(30*60) = 5.6Kbytes/s = 45 kbps.
Of course the actual data size is larger, taking the confirmation frame and other overhead into consideration. But anyway this seems not a problem, because the effective data rate could be 100kbps according to TI MAC introduction (www.ti.com/.../timac


However, I think the data should be sent in CFP period. There are two things to consider here:
(a) CFP can’t occupy all slots in a superframe. There is a minimum requirements for CAP slots. So not all slots could be used by AP to transfer data to ED.
(b) There could be up to 7 GTS in one superframe. So AP could sends data to up to 7 ED in one superframe.
Considering the above two restrictions, I am afraid the throughput might not be able to meet my requirements (45 kbps pure downlink data). If it could not , I think this requirement item could be loosened.
If CAP is used to transfer data from AP to ED, each ED needs to query from AP to check if there is data for it in each superframe, and I think there will be a lot of conflicts and failures.


4. When some events happens in the ED, such a key press by the user, ED needs to send the event notification to AP immediately within 1 second.
Suppose the superframe length is 1 minute and many of the slots are used for CFP, the ED could only send data to ED in the CAP slots of next superframe. That is, the maximum delay time might be 1 minute (or at least several tens of seconds) in worst cases. So How to send it within 1 second?


5. Each ED needs to send a heart beat message to AP indicating it is still alive. The period is around 10 minutes. I think this should not be a problem.


6. Using two CR2450 (supposing the total battery volume is 1000mAH), and one ED needs to receive 5k bytes 10 time per day, ED could work for more than 3 years. I think this might be quite rigorous. The following aspects may need consideration:
(a) Each ED needs to monitor each beacon slot to see if there is data for it.
(b) There might be a lot of conflicts when ED tries to transmit, since there are so many EDs in an AP.
(c) When frequency hopping or other anti-interference mechanism is adopted, ED may need to resync with AP. Do ED need to start searching AP from scratch when AP uses a new channel? Considering there might be a lot of interference in 2.4G, I think this also need be taken into consideration.

  • Hi,
    conceptually yes, the SW is the right one. This being said, the requirements are pretty harsh, in terms of number of nodes as well as throughput. For a beacon mode system, a more realistic target is around 5 to 10 kbit/sec. Please note TI-15.4 MAC does not support GTS. Yes data should be sent during the CFP.
    In order to respect the 1 second latency for the vent from ED to be sent to the AP, then the beacon order should be sent to 1 sec max.
    yes 5. is not a problem.
    6. seems also very challenging with the traffic profile indicated in this thread.
    One option for the interference on 2.4 Ghz is to use a TI-15.4 MAC over sub-1 GHz. We are about to release TI-15.4 MAC 2.0 (with supports for 868 MHz and 915 MHz band) and this can be an option for the customer.
    Thanks,
    TheDarkSide
  • TheDarkSide, thank you for you answer. I still have some questions:

    1. If superframe length is set to 1 sec, ED will wakeup every 1 second to check if there is data available. I think it's too frequent that battery will be exhausted very soon. I heared TI MAC will support 802.15.4 e/g. So does the new standard solve this issue (By using long superframe length, can it also have the ability to send data from ED to AP within short time when necessary?)
    2. One superframe always has 16 slots. I think one slot can't be shared by two transactions at the same time. That is, if one slots is used by ED0 and AP to exchange data, it won't be able to be used by ED1 and AP. If this is true, there can be maximum 15 transactions between 15 EDs and AP within one superframe.
    3. Considering the interference, in the long run, finally there will a lot of interference in 868/915, as well as 2.4G, because they are also license free, and there will be more and more devices running on these bands. It will be more and more crowded, until it could not be used at all. In a high interference environment, it seems very difficult to solve the conflict between throughtput/reliability/latency and power comsuption.

  • 1) If you use beacon mode, there's no other option. 15.4 e/g is not solving this. The latency is the beacon interval for data being sent from the nodes to the AP. If the mode is non-beacon, then you can send data to the AP as soon as you want. But then you have the reverse problem of the latency on the direction down from AP to the node.

    2) yes correct.

    3) I don't agree with this statement. on 2.4 GHz you have in-band interference from the devices in the same network, but you also have interference from other technologies (Wi-Fi, Bluetooth, etc).

    Thanks,
    TheDarkSide

  • Hi, TheDarkSide

    868/915M is not dedicated for 802.15.4, so other devices that are not compatible with 802.15.4 could also run in these bands. So a device using 802.15.4 techonology could also suffer interference from devices from other tech. This is not different from 2.4G.

  • You will never find an un-licensed band (like 2.4 or 868/915) where there's only one technology that operates in it.

    This being said, ISM un-licensed band (like 868 in EU and 915 in US) are at least regulated from an emission standpoint.

    TheDarkSide 

  • Thanks, TheDarkSide. I think the conclusion is that TI MAC is not suitable for my application.