change chunk selection strategy to meet Peerstreamer one
change to peerstreamer type of peer/chunk selection
Add per-edge-events feature to SSSim.
This feature make the simulator acts on per-edge basis rather than on per-node basis.It only works with the event-driven mode.
Add gurobi version for the WuLi optimization algorithm algorithm
After installing the gurobi library in the system, the file "gurobi.sh" should be modified accordingly and sourced before compilation.http://www.gurobi.com/
Fix bug on Wu-Li algorithm, source output capacity
fix bug on source output capacity. In SSSim source always outputs one chunk per chunktime
Fix neighbourhood limit bug in Wu-Li optimization
refactor Makefile due to a compilation error with lpsolve library
fix bug on the node output limit
Refactor with bug fix on A_ij
Refactor inner representation of ajacency matrices
The matrix element A_ij indicates the edge from node j to node i
refactor Wu-Li implementation taking into account the peculiar implementation of transition matrix in SSSim
The element A_ij indicates the edge from node j to node i
Merge Wu-Li algorithm integration feature
add Wu and Li optimization algorithm.
The functionality is activated by loading a graph file with the option ",wuli"
Add per-link events feature
add per-link events.
Add a new flag (-E) to turn on the per-link event. This implies the event-driven mode (-e).Each peer i sends an offer to its neighbor j every offer_interval times probability of j seconds, deterministically.
add .test and .edges to gitignore
add optimization module derived from the paper "Optimal Rate Allocation in Overlay Content Distribution" - actual (centralized) algorithm- lpsolve optimization library as a submodule- test case (really minimal due to the difficulty to solve the optimization problem by hand)
add helper functions to the matrix modules- create a zero matrix- dump a matrix into a .edge file
fix usage print alignment
add auto-convergence feature; simulation stops after delay convergence is reached, a chunk analysis window size can be specified
merge the delay_log feature, solve conflicts
fix test case missing global variable and window analysis size in non-converging mode
Change chunk selection algorithm
add delay logging functionality
Redirect warnings to stdout
redirect Warnings to stdout
Merge branch 'auto_convergence' into next
Conflicts: sssim.c stats.c td.c
add windowed analysis modeadd tests for convergence analysis function
make possible to analyze chunk data on blocks, evaluating the last available.
add auto convergence feature
this feature allow the specification a parameter of convergence on the average receiving delay and the termination of the simulation
add net load factor feature
add network load factor option.
This feature enables to specify the maximum stress on the network during streaming.Default is netload=1, so the sum of chunks sent in one chunk time is at maximum equal to the number of peers.
Add peerstreamer behaviour parameters
Add data logging functionalities
merge graph_matrix branch and fix conflicts.
Add event driven module
Enables a more fine control on the simulation
modify log functionalities by analyse only chunks at steady state
analyze only the last fraction of the chunks in order to have results at steady state
bug fix for vector shrinking (related to offers).
remove annoying tracing
prevent trace option to log scheduling related to overlay monitoring
fix bug in matrix_shrink function.bug affects the shrinking of vectors (one-dimensional matrix)
fix important bug related to peer pointers
add important annotation.reminds that the buffer length defines the playout delay and hence if a chunk can be offered.
1 2 Next » (1-50/86) Per page: 25, 50, 100
Also available in: Atom