Low-Power RF ZigBee® Software & IEEE 802.15.4

Re: How do I implement Z-Stack to another mcu with CC2420?

PME_BLUM — Tue, 05 Aug 2008 16:00:45 GMT

My apologies but at this time we do not provide source code to the ZigBee stack. We highly encourage you to use this community for support and appreciate your understanding as we will do our best to support you moving forward.

Re: How do I implement Z-Stack to another mcu with CC2420?

JP — Tue, 05 Aug 2008 01:36:52 GMT

Hi, PME_BLUM !

Just thinking of it : could it be acceptable to support academic research with source code or hardware details? With all the legal restrictions usually encoutered : NDA, not for resell, not for anything else than the cheap labour made by tired grad students ?

I say that because it is difficult to work with a precise point of interest related to the TI implementation (soft or hard) without support, and your representatives are not always helpful with the extend we would like (sorry ... not their fault). I wish this forum could help but a little chunk of code, or tips from the developpers could greatly improve the support. And this forum is the best place to do it.

Anyway ... just asking (could not hurt).

Many thanks !

Re: How do I implement Z-Stack to another mcu with CC2420?

yangseababy — Fri, 18 Jul 2008 01:26:57 GMT

madman

Re: How do I implement Z-Stack to another mcu with CC2420?

Rock the world — Thu, 17 Jul 2008 08:45:16 GMT

要多国语言并用，呵呵~毕竟老外设计的东西，他们的应用也比我们国内成熟很多

RF技术目前还不成熟，需要进一步发展！Ti开了这个官方论坛，挺不错的

应该可以认识不少国外的高手！

另外协议我们现在自己做，这样移植比较方便

just enjoy it!good luck!

Re: How do I implement Z-Stack to another mcu with CC2420?

wangshuai07 — Thu, 17 Jul 2008 08:05:44 GMT

不知道来这个论坛的有多少能看懂汉语！先汗一个！

z-stack1.4.3现在只支持基于430X核心的cpu，在移植到相同核心的cpu时，比较容易；但要移植到基于430核心的cpu，那是不可能的！

两种核心的编译链接是不一样的！从这点上看，z-stack做的不好！基于430X核心的cpu太贵了！！

Re: How do I implement Z-Stack to another mcu with CC2420?

PME_BLUM — Thu, 10 Jul 2008 15:35:30 GMT

JP,

The discussion has pluses and minuses on both sides, and at this point we will continue to release Z-Stack in object form providing source for the application support layer, cluster libraries, sample applications, low level HAL, and other drivers. This is great discussion and we will continue to revisit this decision in the future to determine whether or not it makes sense at some point to open up the entire stack in source.

Thanks for the great discussion,

Re: How do I implement Z-Stack to another mcu with CC2420?

JP — Thu, 10 Jul 2008 01:06:16 GMT

The comment of Daimon is the reply I would have made.

In another hand, I am wrong or if we develop an application we will certainly need to pass through a lot of certifications and the zigbee one is a tradeof based on if we WANT to change the stack (because the application needs it) or also because the application had to modify in some degree the stack.

Moreover, I know that segmentation of the code in black boxes is great sometimes, and normally we don't need to touch these black boxes but as mentioned, if we want to change hardware and that the communication be optimized or if we want to tune the software independtly of the stack, we sometimes need to know how it is configured, and have the possibility to change some parts of the stack.

And finally, if TI would only support the compiled & certified stack, that would be ok, but if they release former version i.e. as open source, that would be awsome ! I understand also their need to be competitive facing other specialists as Meshnetics and others. But I continue to think that an officially unsupported, non-certified open source would only be a plus versus their concurents. Choice is what will make the difference. Anyway, as I said ... the open source software will be bind to the TI chips so if a developper want to adapt the source to other non-TI, their will be a lot of work to do and probably the choice will stay to the TI part. And anyway the bottom line is that TI will sell silicon. In fact, I am probably too far from the industry economic perspectives as I am working with a research group (and that's also the cause of my interest for open-source).

Another point I think about is why open-sourcing 802.15.4 and not the upper layer ?

Re: How do I implement Z-Stack to another mcu with CC2420?

Daimon — Wed, 09 Jul 2008 09:51:12 GMT

PME_BLUM,

I agree in the point that almost all TI's customers are just interested in adding a application over the Z-stack, therefore it's better not to open up the code. However as a different project, which can't be supported by TI and in this case the compilant certification is the customer's problem, the code can be open up. In that situation TI will have a compilant platform as now and in addition the advantatges of a open project. For example, I'm developing a new product based on CC2430 and Z-stack, but I want to add an external RAM memory which my application will use without problem, but the Z-Stack ( I'm specially interested in the routing mecanism) won't be able to use it. I know that I won't have support and I will have to pass the stack certification process before the product certification one( maybe the product is not certified after all), but that improvement is very interesting for my application.

I understand TI point of view, so I won't insist, but it's an intersting conversation.

Re: How do I implement Z-Stack to another mcu with CC2420?

PME_BLUM — Tue, 08 Jul 2008 22:42:09 GMT

Daimon,

Your feedback is definitely something requested often, but unfortunately the situation is not as simple as opening up the code. First, going through the product certificaiton process is different then the stack certification process, and while TI can currently make the statement that we provide our customers with a ZigBee compliant platform on which to build a product, we could no longer make this statement if we provided source code that a customer could potentially modify. If even a single line of code is modified then technically the customer would need to re-certify this. We realize this is not impossible, just something that is unlikely given that most customers want to focus there efforts on the application and not have to worry about the networking aspects of ZigBee (hence Z-Accel, but a completely separate topic). Anyway that said, another issue is that as you probably know the ZigBee stack, most specifically the networking portion of it, is quite complex and there is significant resources in development and testing of this code to make it as bug free and performance enhanced as possible. Opening up the code does offer the benefits of potentially allowing for customer improvement, but would likely result in customers going down paths that we simply don't have the bandwidth to support. As mentioned by several others, there are ZigBee stacks out there in the open source community which are available. However realizing the time and effort that has gone into fixing bugs and stress testing the system with 10's to even 100's of nodes, I caution most about building a product on one of these open source solutions without performing this extensive testing on their own. Anyway great conversation nonetheless. If opening up the source would make our customers more successful, I believe we would do this, but when taking the more global view of the ZigBee market and it seems that the best bet to help our customers be successful is to keep them focused on their application and hide the details of networking.

Re: How do I implement Z-Stack to another mcu with CC2420?

Daimon — Tue, 08 Jul 2008 14:39:51 GMT

Hi,

I agree with you in the main point, but as you can modify part of the ZDO or the OSAL, which are not really the Zigbee stack itself, why you can't modify the other files?. Of course TI won't have to provide support to modified stacks.

On the one side If the code is opened people can contribute with improvements or suggestions, which TI can add to Z-stack, or not.

On the other side who wants to modify the stack has to deal with the Zigbee compilance testing ( In adition, Everybody who wants the Zigbee logo for its product has to pass the Zigbee certification product program), but at least we can choose.

It depends on each one interests.

Re: How do I implement Z-Stack to another mcu with CC2420?

LPRF Rocks the World — Tue, 08 Jul 2008 11:08:37 GMT

Hi,

That is right. Every new build for a specific platform will require compliance testing to obtain the status of a ZCP - ZigBee Compliant Platform.

http://www.zigbee.org/en/certification/index.asp

LPWRocks

Re: How do I implement Z-Stack to another mcu with CC2420?

emmsys — Mon, 07 Jul 2008 13:40:49 GMT

Hi,

Lots of great information! I'm just playing devil's advocate now (and I don't work for TI!) but from TI's perspective, and given how new ZigBee is, they can provide better support if the Z-stack runs on a limited number of their own platforms. If they open up the Z-stack to everyone then it won't be possible to provide support. Also, these stacks are certified as ZigBee compliant platforms. If you move the Z-stack to other platforms than the whole package needs to be re-certified. Lots of interesting pros/cons for all of this!

Emmsys

Re: How do I implement Z-Stack to another mcu with CC2420?

Daimon — Mon, 07 Jul 2008 09:54:37 GMT

Hi all,

Related with Zigbee, Open sources and TinyOS. There is another IEEE802.15.4 / Zigbee Stack written in nesC for TinyOS, it is called Open-ZB and is developed by a research group called IPP-Hurray.

If you want more information, its website is: http://www.open-zb.net/

You can get their stack in the download area. It is not complete, but it works.

Daimon

Re: How do I implement Z-Stack to another mcu with CC2420?

LPRF Rocks the World — Mon, 07 Jul 2008 08:25:03 GMT

Hi JP,

You have some great points. Open source commnuities are great, such as the TinyOS coming out of Berkeley, http://www.tinyos.net/

In terms of ZigBee and open source communities I am only aware of one. They started out when ZigBee 2004 came to the market. This specification is of course now obsolete. I do not know the current status of this project. This community contacted us three years ago. See below:

From the OpenBee community that contact us back in 2005:

"We are two engineers working at the University of Applied Sciences Valais, Sion, Switzerland (http://www.hevs.ch). We have begun the development of a ZigBee compatible stack under GNU/GPL license. The stack will be microprocessor, radio transceiver and operating system independant.

At the moment, we are working on the structure of the OpenBee Stack. For this, we use the UML tool Rhapsody from I-logix (http://www.ilogix.com), to speed up the design and shorten the implementation time. During this phase we will not check in any code into the sourceforge CVS repository, because the code is generated out of the UML description (C code is generated). The proposed structure of the code will be regularly posted on http://www.openbee.org and your comments are welcome (release 1.x already available in the member area). Once the architecture of the stack is defined, function declarations for the different parts (MAC, NWK, APS, AF, security, route discovery, etc.) will be available and the collaborative development can begin. If you announce your interest for a specific part, we'll provide you with the corresponding functions declarations.

Rhapsody design files will be made available both as UML graphs and as Rhapsody source file (Rhapsody licence required!). If you do not intent, or could not use rhapasody as development tool, we can take your source files and reverse engineer it into the Rhapsody project. This means you can work with your favorite IDE.

Don't hesitate to propose your ideas. To keep up with the latest news on OpenBee, just register on the OpenBee website www.openbee.org This will give you access to the member area of the site and will add your email address to the news mailing list. "

Hope this helps.

LPWRocks

Re: How do I implement Z-Stack to another mcu with CC2420?

JP — Fri, 04 Jul 2008 15:03:32 GMT

Hi all,I am aware of all that. I know that TI, as now would not be realistically be able to support every kind of microcontroller available. In a matter of facts, my suggestion was more for the discussion than just a real current need (as I currently opted for the CC2431 as a replacement for the CC2420 and it is working great as a coprpocessor). But I really think that would be great for every one if the Z-stack could be open to the public. "Great" because form my experience, one of the big challenge with Zigbee, is the difficulty to use libraries. I think that since TI is a silicon manufacturer, they will not suffer from opening the source code, in the contrary that will (maybe) attract more customers to use the chip of an open solution and they will sell more silicon.

Up to now, I worked with a Silicon Labs MCU and a CC2431 as a coprocessor especially for the location engine; my remark was more for the matter of discussion than a real need (but I am interested with the alternatives provided above). Both Silabs and TI provide libraries to implement Zigbee. I like the MSP430 also, that is a great product, but is not powerful enough for the application. At the beginning of the project, I was obligated to infer/deduct the comportment of the code to be able to customize it (especially with the Silabs library and the crossbar setting). That was really ... painful. And I ended to realize some parts of the Zigbee Stack on both MCU! If the code was released, I could just read the HAL part, adapt the configuration to my needs and be able to code cleanly my application on top of the Zigbee stack for more configuration than just MSP430+CC2420. I could also as suggested by Emmsys just let the code in the CC2431, but the project I am currently working on needs to interface closely with the location engine and the Zigbee.

Moreover, the customer will be more attracted to an open and well supported software (like all the Linux and others), and I think it is a current drawback of Zigbee and restraining the adoption of this standard.

That’s why I think TI could only be winner of supporting other solutions, if they continue to provide interesting hardware (distributed location engine, low power transceiver ...) as Chipcon and now TI used to do.

What is the point of vieww of all of you ?