Lightweight Internet Traffic Classification - A Subject Based Solution with Word Embeddings

Internet traffic classification is a relevant and mature research field, anyway of growing importance and with still open technical challenges, also due to the pervasive presence of Internet-connected devices into everyday life. We claim the need for innovative traffic classification solutions capable of being lightweight, of adopting a domain-based approach, of not only concentrating on application-level protocol categorization but also classifying Internet traffic by subject. To this purpose, this paper originally proposes a classification solution that leverages domain name information extracted from IPFIX summaries, DNS logs, and DHCP leases, with the possibility to be applied to any kind of traffic. Our proposed solution is based on an extension of Word2vec unsupervised learning techniques running on a specialized Apache Spark cluster. In particular, learning techniques are leveraged to generate word-embeddings from a mixed dataset composed by domain names and natural language corpuses in a lightweight way and with general applicability. The paper also reports lessons learnt from our implementation and deployment experience that demonstrates that our solution can process 5500 IPFIX summaries per second on an Apache Spark cluster with 1 slave instance in Amazon EC2 at a cost of $ 3860 year. Reported experimental results about Precision, Recall, F-Measure, Accuracy, and Cohen's Kappa show the feasibility and effectiveness of the proposal. The experiments prove that words contained in domain names do have a relation with the kind of traffic directed towards them, therefore using specifically trained word embeddings we are able to classify them in customizable categories. We also show that training word embeddings on larger natural language corpuses leads improvements in terms of precision up to 180%.

Identity construction through phonetic crossing among young Capetonian gang members

Uncountable gangs operate in post-Apartheid South Africa, particularly in greater Cape Town, competing over turf and controlling the drug trade. Consequently, gang violence is rife in Western Cape and especially widespread in urban areas. In this paper young Capetonians’ narratives of gang violence are analyzed. In the narratives of attacks on Black or White South Africans by Coloured gang members, the Coloured narrators make use of their victims’ varieties of English, more precisely, of phonetic features. Hence, the aggressors do language crossing towards their targets when narrating their feats. Rampton (1995a:485) considers language crossing a ‘code alternation by people who are not accepted members of the group associated with the second language that they are using (code switching into varieties that are not generally thought to belong to them)’. This switching involves a transgression of social or ethnic boundaries that allows the young gangsters to construct, negotiate, uphold and manage their social identities, as language still functions as an utterly important identity marker in post-Apartheid South Africa.

Putting Up a Big Front: Car Design and Size Affect Road-Crossing Behaviour

Previous research suggests that people tend to see faces in car fronts and that they attribute personality characteristics to car faces. In the present study we investigated whether car design influences pedestrian road-crossing behaviour. An immersive virtual reality environment with a zebra crossing scenario was used to determine a) whether the minimum accepted distance for crossing the street is larger for cars with a dominant appearance than for cars with a friendly appearance and b) whether the speed of dominant-looking cars is overestimated as compared to friendly-looking cars. Participants completed both tasks while either standing on the pavement or on the centre island. We found that people started to cross the road later in front of friendly-looking low-power cars compared to dominant-looking high-power cars, but only if the cars were relatively large in size. For small cars we found no effect of power. The speed of smaller cars was estimated to be higher compared to large cars (size-speed bias). Furthermore, there was an effect of starting position: From the centre island, participants entered the road significantly later (i. e. closer to the approaching car) and left the road later than when starting from the pavement. Similarly, the speed of the cars was estimated significantly lower when standing on the centre island compared to the pavement. To our knowledge, this is the first study to show that car fronts elicit responses on a behavioural level.

Modeling of Self-similar Traffic

Modeling of self-similar traffic is performed for the queuing system of G/M/1/K type using Weibull distribution. To study the self-similar traffic the simulation model is developed by using SIMULINK software package in MATLAB environment. Approximation of self-similar traffic on the basis of spline functions. Modeling self-similar traffic is carried outfor QS of W/M/1/K type using the Weibull distribution. Initial data are: the value of Hurst parameter H=0,65, the shape parameter of the distribution curve α≈0,7 and distribution parameter β≈0,0099. Considering that the self-similar traffic is characterized by the presence of "splashes" and long-termdependence between the moments of requests arrival in this study under given initial data it is reasonable to use linear interpolation splines.

Students crossing State St. in front of Michigan Union. Taken for the Michigan Alumnus, April, 1947

On negative sleeve: A suggestion of warmer days ahead for winter-weary Michiganders is contained in this early spring picture, taken from the Michigan Union looking across to Alumni Memorial Hall just as students leave classes at the noon hour

Additional traffic assignment options for the TASSIM model. Phase I. Final report.

Transportation Department, Washington, D.C.

Alcohol highway-traffic safety workshop for the judiciary.

National Highway Traffic Safety Administration, Washington, D.C.

Preliminary human factors guidelines for traffic management centers.

Federal Highway Administration, ITS Joint Program Office, Washington, D. C.

Evaluation of traffic control devices for rural high-speed maintenance work zones.

Texas Department of Transportation, Research and Technology Transfer Office, Austin