[Ns-developers] GSoC 2010: WPAN models: 802.15.4 implementation for ns-3

[JuanLu]

Hello:

I am Juan Luis Font, PhD student from the University of Seville, Spain.

As I mentioned in a former e-mail, I am interested in develop some models
related with Wireless Personal Area Networks (WPAN). Specifically I want to
focus on IEEE 802.15.4 technology, used in the context of WPAN due to its
simplicity and low power requirements. I have already some theoretical
knowledge about this topic and I am getting familiar with ns-3 class
hierarchy in order to integrate the new models into the source code
structure.

I would like to describe briefly some of the characteristics and
functionalities of a possible 802.15.4 model for ns-3. Due to the nature of
the Standard, the model would cover only the channel, Physical layer and
Medium Access Control sublayer (plus SSC sublayer, described later).

In general lines, it would be very useful that the 802.15.4 devices could be
associated with ns-3 nodes as other wireless devices already do, as well as
taking advantage of the existing mobility, loss and delay propagation
models. As far as I know, following the ns-3 coding recommendations and
defined architecture would provide this characteristic without much effort.
If necessary, some extra mobility and propagation models could be added in
order to match specific needs related with WPAN context.

Wireless channel:
Initially, I would create a new class that would inherit from the current
generic wifi channel. No new fancy features are planned (for now).


Frames:
Attending to the 802.15.4 Standard, there are two main types of frame, for
physical and MAC layer. The last one is subdivided into 4 different type of
frame.
  - PHY
  - MAC
      - Beacon
      - Data
      - ACK
      - MAC command


PHY layer:
This layer deals with characteristics of physic channel. All the frequency
bands would be contemplated, independently of their geographical
availability. This part of the model would include the necessary methods to
deal with:
  - Frequency bands
      - 800 MHz
      - 900 MHz
      - 2'4 GHz
  - Data rates
  - Radio characteristics
  - PHY services
      - Data services
      - Management Services
      - PAN information Base Management Primitives
  - Energy detection


MAC sublayer:
This sublayer provides access control to a shared channel and reliable data
delivery. Together with the LLC layer, they comprise the data link layer. An
auxiliary sublayer is defined in the Standard in order to connect this MAC
sublayer with IEEE 802.2 LLC, named Service Specific Convergence Sublayer
(SSCS).

  - Topologies: the IEEE 802.15.3 MAC layer allows the creation of two
different types of topologies: star and peer-to-peer
  - MAC services
      - Data service
      - Management service
  - MAC PAN Information base management primitives
  - Radio channel scanning
  - Association an disassociation control
  - Guaranteed time slot management
  - Orphan device management
  - Synchronization control
  - Beacon Management
  - Beaconless synchronization
  - Security services (optional, open to discussion)

Although  topologies are not part of the Standard, it defines the tools that
allow their creation, specifically star and peer-to-peer ones.
      - Star: relies on a network coordinator and beaconed mode
      - Peer-to-peer: allows more complex organizations, special interest in
cluster tree topology.

In order to test the above models, it is especially interesting for me the
creation of star and cluster tree topologies. The study of propagation of
beacons and node synchronization in complex topologies other than star is
one of my priorities related with these models.

I would be interested in coding a power consumption model, but for now I am
going to leave it off this first GSOC proposal. I have checked other
students' ones and some of them are interested in power consumption models
too, so I will follow the development of those discussions.

Most of the above points follow the definition of the IEEE 802.15.4
Standard, so comments and opinions about which of them are worth
implementing (or not) are welcome. I may be doing an approximation too close
to real systems so if you think I should increase the level of abstraction.

On the other hand, I cannot evaluate accurately the overall complexity of
this task in terms of amount of code and needed time. If you estimate this
would be a very simple proposal that would not meet GSOC expectations let me
know and I will do my best in order to add some extra goals related with
this field.

Best regards.

-- 
JuanLu