Benchmarks :
"Location Areas Planning Optimization In Mobile Networks"

Contact: laidi.foughali@lifl.fr

The principle of the network benchmark generation is as follows. We generate for each cell of the network the cartesian coordinates corresponding to the physical site of its associated BS (Base Station). As a preliminary, we decide of the global shape which will take a network. We triangulate the set of points by using the Delaunay triangulation. Thus, we obtain a planar graph where each node represents a cell and each edge represents potential mobility between two neighbor cells. The benchmarks can be formed from 300 or 600 cells with 4, 8 and 16 BSCs having the shape of the Fig. 1. The BSCs groups could have the shape of the Fig. 2 (square) or of the Fig. 3. (rectangular)
Example of planar graph with 600 cells
Fig. 1
 Square hull convex
Fig. 2
 Reactangular hull convex
Fig. 3


Once the planar graph is obtained, we define the weights of the edges and the nodes by generating random values included between a lower limit and an upper limit relative to the taffic, TRX, SDCCH, MT-BHCA/PAG and the location updates. The various limits are summarized in table I. The values of constraints necessary in the optimization are summarized in the table II.

Variable Lower limit Upper limit
Traffic 6 10
TRX 2 3
SDDCH 2 2
MT-BHCA/PAG 250 300
Location updates 0 100
Table I: Limit values (nodes weights)

Entity Capacity
Traffic 6
Network 7 (+/-3) MSCs
MSC 350000 Subscribers
BS 3 Cells Max.
BSC 30 BSs
LA 60 Cells
Traffic for MSC 15000
Traffic for BSC 3000
Traffic for LA 1000
TRX for MSC 1890
TRX for BSC 630
TRX for LA 300
SDCCH for MSC 1260
SDCCH for BSC 600
SDCCH for LA 400
MT-BHCA/PAG for MSC 175000
MT-BHCA/PAG for BSC 100000
MT-BHCA/PAG for LA 60000
Cells that can handle a MSC 3000
Cells that can handle a BSC 100 (Direct)
Cells that can handle a BSC 400 (Indirect)
Cells that can handle a LA 200
Table II: Capacities of the network entities

Every benchmark contains 6 files ('lapo.msc', 'lapo.bsc', 'lapo.cell', 'lapo.link', 'lapo.neighb' and 'lapo.param'). Each one describes the characteristics and constraints of an entity of the network to be optimized :
  1. The 'lapo.msc' record contains :
    • Identity
    • Geographical coordinates
    • Max. Traffic (Erlang)
    • Max. TRXs number (Signaling and communication)
    • Max. SDDCHs number (Signaling)
    • Max. MT-BHCA/PAG (Signal)
    • Max. Managed cells
  2. The 'lapo.bsc' record contains :
    • Identity
    • Geographical coordinates
    • Parent msc
    • Max. Traffic (Erlang)
    • Max. TRXs number (Signaling and communication)
    • Max. SDDCHs number (Signaling)
    • Max. MT-BHCA/PAG (Signal)
    • Max. Direct managed cells
    • Max. Indirect managed cells
  3. The 'lapo.cell' record contains :
    • Identity
    • Geographical coordinates
    • Parent bsc
    • Avg. Traffic (Erlang)
    • Avg. TRXs number (Signaling and communication)
    • Avg. SDDCHs number (Signaling)
    • Avg. MT-BHCA/PAG (Signal)
  4. The 'lapo.neighb' record contains a binary vector where each element indicates if the corresponding cells are neighbors.
  5. The 'lapo.link' record contains a binary vector where each element indicates the location updates of the corresponding cells.
  6. The last file 'lapo.param' is similar to all benchmarks. It describes the location area constraints:

Benchmarks list:
  1. 300C4Bs
  2. 300C4Br
  3. 600C8Bs
  4. 600C8Br
  5. 600C16Bs
  6. 600C16Br