|
On the web!
|
10/10/2008
|
| |
Web enabled devices are springing up everywhere and it is seen as an advantage if they can be controlled, monitored or updated remotely. So it’s no surprise that many people feel their latest product needs to be connected to ‘the internet’.
Having decided to use some form of internet connectivity, the biggest problem is finding a way to actually make the link from a small embedded application to one that can communicate with large servers on the other side of the world. Whilst there are commercially available TCP/IP stacks, they may cost a lot, as well as requiring much time and effort to integrate with the product.
The stack usually requires an operating system in order to run effectively. Packets of data can arrive on the Ethernet cable at any time and there are all manner of responses that might need to be sent in return as the packet is decoded by the different layers of the software. On the other side, transmitting data is not just a case of handing it over to the driver and saying ‘go’. With the reliable protocol TCP, the operating system has to monitor responses from the device at the other end and be prepared to retransmit the data until it knows that it has been received successfully. This clearly takes cpu power and memory, which has to be managed by the software. In most cases, this is why operating systems are brought into the equation because it is easier to split each of these different functions into separate tasks which can run semi autonomously.
It is at this point that the networking aspect of the project is often put to one side because it is seen as being too complex. There is new software to write or buy and new APIs to learn.
|
| |
Author
Sean Cochrane
|
| |
| |
|
This material is protected by Findlay Publications copyright 2008.
See Terms and Conditions.
One-off usage is permitted but bulk copying is not.
For multiple copies contact the sales team.
|
| Email this article |
| |
|
|
|
|