Automatic premises network design using standardised building data

1
Automatic premises network design using
standardised building data

• Haruhisa Nozue, Takashi Goto, Fumihiko Ito
• NTT Access Network Service Systems Laboratories,
• NTT Corp.

2
Purpose

• Support premises network design
• with automatising physical wiring design.
• To enjoy the convenience of network appliances,
  we need an appropriately designed network.

3
Background in the near future...

• Many users may be forced to construct their own
  premises networks by themselves.
• Because of the variety of user demands and
  environments.
• It will be difficult for an unexperienced user to
  construct an appropriate network.
• Because of the variety and the complexity of
  devices.

4
Background emerging technology

• Spread of standards for building plans with
  semantic data.
• Not only drawings but objects and their
  relations.
• E.g. Industry Foundation Classes (IFC).

• Storey1
• Room1
• Openings
• Door1, door2, ...
• Boundary
• Wall1, wall 2, ...
• Room2 ...
• Storey2 ...

5
Goal of this work

• Automatically design the suitable physical wiring
  using a building plan.
• Input
• Users Building plan with semantic data.
• Positions of devices and interfaces already
  assigned.
• output
• Most appropriate wiring design.
• i.e. positions of devices and paths of cables.

6
Basic ideas

• Formulate network design into a graph
  optimisation problem.
• Based on a common sense, generate a template
  (weighted graph) for network design from a
  building plan.

• e.g.
• Devices and cables should not be located far from
  walls.
• We prefer to lay cables along walls rather than
  in front of doors, windows and closets.

7
Outline of our method

• Generate a wiring template graph for the building
  from the building plan.
• Assign costs to each edge of the graph.
• Derive and obtain an optimum subgraph
  corresponding to a suitable network design.

8
Step 1 generate a template graph
IFC file
Template graph
9
Step 2 Assign costs to each edge

• The longer, the higher cost.
• Huge costs to
• Paths between different rooms (c4, c7, c12, ).
• Paths across the front of openings (e.g. c14ltc16).

10
Step 3 get an optimum subgraph representing the
suitable network.

• The connected subgraph which has the minimum cost
  and includes all assigned points is the most
  suitable network design.

11
Conclusion and future work

• We have proposed the basic procedures of premises
  network design using a building plan with
  semantic data.
• To deal with more realistic situations more
  appropriately
• Multiple domains
• Detailed methods for extracting objects in
  complex plans
• Cost functions

12
Detailed procedure - 1 derive the room topology
of the user building
13
Detailed procedure - 2 abstract candidates for
device positions

• e.g.
• Corners.
• Wall sockets and network interfaces.
• Where a device is already located.
• Sides of openings.

14
Detailed procedure - 3 generate wiring templates
for each room

• Along room boundaries.

15
Detailed procedure - 4 generate a wiring
template for a whole building

• Combine small graphs with paths of room-to-room
  wiring.