Detta jobb har gått ut Se liknande jobb
Master's thesis Media Delivery Networks
The great popularity of smartphones, tablets, and laptops all wirelessly connected to Internet has caused an exponential traffic increase over cellular networks. The continuous increase of multimedia information accessed by wireless devices is overloading the core network that connects these devices to the servers providing the desired content. Such an increase of data traffic has caused an immediate need for offloading traffic for optimum performance of both voice and data services. Services such as real-time video downloading are potentially being disrupted by the delays to access the content.
Caching is a common way to alleviate this problem and offload the network. Here, the relevant content is pre-stored in nodes or access points closer to the end users. However, while the static caching mechanism is well understood for single caches, it becomes much more complex for dynamic multi-caching systems on large scales, where an individual caches may be populated by using both remote server content, and cooperatively other caches contents, depending on the average and real time request dynamics. How to automatically populate the catches in these systems is largely an open problem.
In this exciting master thesis project, we consider the problem of offloading the network by optimally placing the content to access points so as to ensure service availability for the end users. First, a comparative studying of existing algorithms will be carried out. Then, pre-caching mechanisms and on-demand mechanisms will be investigated.
In the pre-caching mechanisms, access points will be pre-populated based on learning from past history. The population should be done by predicting the request patterns of the future requests and should be based on the available sequences from the past. This will allow populating the caches by a streaming toward the nodes in right moments when the network is not congested. In the on-demand caching, distributed algorithm where the access points cooperatively fetch data both from other catches and serves based on real-time requests will be investigated.
The challenge is that the streaming should avoid congestion. The streaming is not free, meaning that there will be a cost per unit of information, depending on time and space. Also, the storage at the access points has a cost. Moreover, the caching algorithm will work on Layer 7, where in case there may be additional information coming from probing of lower layers. The underlying network is an overlay network.
The ultimate goal is finding algorithms that place the content in an automatic manner, and not based on static or manual specifications, so to ensure real-time service availability for the end user. The optimal algorithms will work on a network abstraction where the optimal solution depends on the time and place to offer highly performing connectivity.
Master Degree (D,E,Y) or similar.
This is a highly exciting project between the Automatic Control Department of KTH Royal Institute of Technology and Ericsson AB, Stockholm, Sweden.
Those who are interested to work in an interdisciplinary field between mathematical network optimization and content distribution networks and learn new techniques and broaden the knowledge are encouraged to apply. Any Master’s student who has good interest in mathematical optimization, game theory, automatic control and networking is welcome. Electrical engineering, engineering physics, computer sciences students with strong mathematical knowledge are encouraged to contact the persons below.
Starting time: from end of August 2013
Note, to apply, please use the apply button in the ad.
Do NOT apply via the contact persons e-mail addresses!
Logga in med Facebook och LinkedIn så får du se vilka av dina vänner som jobbar här