Reliable Opportunistic Mobile Networks
- AG Waldvogel (Distributed Systems)
|(2011): Prediction quality of contact history in opportunistic networks 2011 IFIP Wireless Days (WD). - IEEE, 2011. - S. 1-3. - ISBN 978-1-4577-2027-7|
Routing in opportunistic networks heavily relies on past behavior of the mobile devices it is formed of to predict their future and thus making routing decisions. While almost every protocol relies on this history, its prediction quality has never been studied in a realistic setting. Using extensive simulations on real traces, we are able to describe for the first time how well predictions can be. Unlike oracle-based prediction comparisons, we do not try to predict a contact, but compare the expected user result, namely message delivery probabilities. The analysis also provides guidance on the importance of multi-path routing and the path diversity required, as well as on the impact on forward error correction on the delivery probability. Our results show that the repetitive nature of path is directly proportional to the mobility extent of the devices and, consequently, history obtained from dense opportunistic networks is reliable.
|(2011): Optimizing message delivery in mobile-opportunistic networks 2011 Baltic Congress on Future Internet and Communications. - IEEE, 2011. - S. 134-141. - ISBN 978-1-4244-8511-6|
The big challenge of routing in opportunistic mobile networks, overlooked by most researchers, is to not only find any path to the destination, but a path that is stable and powerful enough to actually carry the message. Few attempts addressed this problem, all of them under controlled scenarios, avoiding the complexity of real-world connectivity. As a result of our comparison of selected networks under a wide variety of realistic scenarios, we have not only been able to identify and describe favorable traits of protocols, but also necessary relationship of successful MON protocols with QoS routing in wired networks. We present a novel protocol, Nile, that performs both in dense as well as sparse networks. Nile is the first autonomous “controlled Hooding” protocol that keeps the link loads in check, to push replicas only on those paths that are both promising and may sustain more load. It is a multi path protocol that deploys replication based on heuristic for disjoint path calculation. Other protocols' performance, when simulated in real-world traces, highly depends on parameter choice. Nile, however, consistently performs among the top protocols without any external tuning and exerts far less overhead than other replication protocols.
|(2011): Questioning flooding as a routing benchmark in Opportunistic Networks 2011 Baltic Congress on Future Internet and Communications. - IEEE, 2011. - S. 128-133. - ISBN 978-1-4244-8511-6|
Routing in Opportunistic Networks, as a relatively young discipline, still lacks coherent, simple and valid benchmarks. It is customary to use epidemic routing as performance benchmark for Opportunistic Networks. We identify and describe the current simulation practices that do not expose the shortcomings of flooding as an upper bound. In this paper to provide a step towards a routing benchmark, which is flexible, provides results close to an upper bound, is simple to implement, and thus might be a candidate for a common benchmark. This new method called EPO, does not suffer from bottlenecks that limit the performance of epidemic flooding, even when bandwidth is scarce. Our analysis shows that networks are not suffering from that much severe congestion as suggested by flooding and thus giving a better insight to the underlying network.
|(2011): Evaluation and Innovation in Opportunistic Networks|
The presence of an increasing number of mobile devices has prompted the demand to use them for information dispersion through opportunistic networks, which they form coincidently due to their geographic location. Opportunistic networks pose several new challenges to the current transmission protocols as they are not only capable of store and forward routing, but also lack the offline routing capability, i.e. source and destination must be connected to the network simultaneously. We can find several opportunistic network protocols in literature, but neither a solid comparison nor a trusted baseline has been presented.<br />In this study, we have analyzed and evaluated eight existing routing algorithms on a common basis in an effort to grasp the strong and weak points of each of them and to see whether it is possible to design a hybrid technique that may take advantage of the strengths of several other techniques. We propose three important criticisms regarding the evaluation of existing routing algorithms.<br />1. Most evaluations restrict themselves to comparing against the two extremes, direct-contact forwarding and flooding.<br />2. Each attempt uses a completely different choice of scenario and simulation parameters.<br />3. Most attempts concentrate on methods to find a path to destination but the reliability of the path cannot be ensured in this paradigm.<br />The findings have revealed that almost all the techniques fail to perform under variable conditions, i.e. bandwidth hungry techniques failed to deliver when bot- tlenecks existed although, they outclassed every other technique where network had sufficient capacity. In contrast, techniques that required good network con- nectivity failed to perform in sparse network. As a result of our comparison of selected networks under a wide variety of realistic scenarios, we have not only been able to identify and describe favorable traits of protocols, but also neces- sary relationships of successful mobile opportunistic network protocols with QoS routing. This study defines a very light weight metric, which not only encapsu- lates the path bandwidth but also maintains a dynamic path ranking by degrading the path efficiency as it suffers from data load. Moreover, this study focuses on the routing algorithm Nile1 that has an adapting capability with the underlying network. It thereby maintains acceptable performance without exhausting the network resources keeping a check on the network “pulse”, i.e. bandwidth. Nile is a multi-path protocol that deploys replication based on heuristic for computing disjoint path . Flooding is considered to be a protocol that can deliver the best performance if its overhead is ignored. Therefore, it is customary to use flooding as a performance benchmark for opportunistic networks. We identify and describe the current sim- ulation practices that do not expose the shortcomings of flooding as an upper bound. We provide a step towards a routing benchmark, which is flexible, pro- vides results close to an upper bound, is simple to implement, and thus might be a candidate for a common benchmark. This new method called EPO2, does not suffer from bottlenecks that limit the performance of epidemic flooding, even when bandwidth is scarce. Our analysis shows that networks are not suffering congestion as suggested by flooding, giving a better insight in the underlying network.<br />Since most of the practical routing protocols rely on history to profile devices, these profiles are used to compute the routes in the network. We found that history is either not able to predict device behavior accurately or history looses the details about the device behavior due to aggregation of metrics. We have therefore analyzed opportunistic network with max-flow to see the throughput of the network. We afterwards compared the outcome of this max-flow with a modified max-flow that uses history for its throughput computations. We have found that it is not possible to obtain an accurate history based max-flow, since it is not easy to find the link in path that can be considered to be responsible if history gives an unreliable path.
|(2008): I Seek for Knowledge : Exploiting Social Properties in Mobile Ad Hoc Networks 2008 1st IFIP Wireless Days. - IEEE, 2008. - S. 1-5. - ISBN 978-1-4244-2828-1|
New social networks are born each day, at a formal conference, at informal social gathering, at family reunions etc. Internet has already been playing an important role in modern way of socialising. But it is still not the optimal way of interaction as one has to be very active updating profiles. With the easy access to mobile devices, modern technologies have now started to adopt to new ways of socialising. mobile devices accompany their users almost all the time, they can record and observe their users behavior as well as gather information about their social circle. Therefore, they can help users to get information from contacts, that they probably not even know. In this paper we put our efforts towards the initial design of an architecture that will sniff for information around the user's surrounding, leveraging useful answers on their demand.
|(2008): Reality-Check for DTN Routing Algorithms 2008 The 28th International Conference on Distributed Computing Systems Workshops. - IEEE, 2008. - S. 204-209. - ISBN 978-0-7695-3173-1|
Many applications of ad-hoc networks include intermittent connectivity. Anyone wishing to implement routing into her delay-tolerant network can select from a wide variation of options, but the choice is hard, as there is no strong comparative evidence to the relative performance of the algorithms. Every paper uses a different setting, mostly far from realistic. In our desire to improve the basis for decisions, we simulated a promising selection of DTN routing algorithms in three vastly different scenarios, all based on publicly available real-world traces. Using our open-source DTN simulator, we compare and analyse 11 routing techniques, then provide explanations for the behaviour and give advice for choosing a suitable mechanism. To our own surprise, the results challenge the conventional wisdom gained from synthetic simulations and poses the question whether the world is ready for DTNs.
|(2008): Nuntifix-Modeling of Delay Tolerant Networks : A Technical Report|
In the past years, many promising routing algorithms for delay-tolerant networks have been described as well as simulated or even implemented, and also evaluated. Anyone wishing to implement routing into delay-tolerant network can select from a wide variation of options, but the choice is hard, as there is no strong comparative evidence to the relative performance of the algorithms. While each algorithm by itself is exciting and adds to our world s knowledge, one important aspect of research is to understand when to do what. As the algorithm evaluations lack a common basis, comparison between the algorithms is currently impossible: First, most evaluations restrict themselves to comparing against the two extremes, namely direct-contact-only forwarding and ﬂooding; second, each attempt uses a completely diﬀerent choice of scenario and simulation parameters. In this report, we analyze and evaluate the currently existing algorithm under the common basis in an eﬀort to grasp the strong and weak points of each of them and to see whether its possible to design a hybrid technique that may take advantage of strengths of many techniques.
|Deutscher Akademischer Austauschdienst||-||Stipendium (HEC/DAAD)||no information|
|Period:||01.04.2006 – 31.03.2009|