We use ndnSIM [5] as our simulation platform. NdnSIM provides several forwarding
strategies including BestRoute which allows only one outgoing face and
SmartFlooding which allows multiple outgoing faces. We use a selected
Rocket-fuel topology AS-1755 [6]. Leaf nodes are randomly assigned as producers
and consumers. Backbone and gateway nodes are assigned as routers. We vary the
percentage of censor routers from 0 to 100 % to represent the effectiveness of
censorship.
Consumers send Interest packets in the first 10 s. The time to start these requests
follows uniform distribution. 30 % of these Interests are with censored names and
the Data packets of the remaining Interests are with censored content. Therefore
when the censorship is effective enough, no files can be successfully received. We
considered both situations of sparse traffic and dense traffic since the significance of
Content Store is not the same in these situations. Each consumer sends out 5 and 60
Interests in the situations of sparse and dense traffic respectively. Each file is
separated into several blocks and we say a file is successfully received if all blocks
have been received. The simulation lasts for 500 s. The percentage of successfully
received files using different forwarding strategies is given in Fig. 1.
Figure 1 shows that with the same percentage of censor routers, multiple outgoing
faces forwarding enables consumers to receive much more files than single
outgoing face forwarding. This indicates that the multiple outgoing faces forwarding
strategy in NDN weakens the effectiveness of censorship and favours the
spread of information.
Figure 1 also indicates that even if SmartFlooding is used, the percentage of
successfully received files decreases with the increase of censor routers. In other