This thesis tackles problems on IEEE 802.11 MAC layer, network layer and application layer, to further push the performance of wireless P2P applications in a holistic way. It contributes to the better understanding and utilization of two major IEEE 802.11 MAC features, frame aggregation and block acknowledgement, to the design and implementation of opportunistic networks on off-the-shelf hardware and proposes a document exchange protocol, including document recommendation.
First, this thesis contributes a measurement study of the A-MPDU frame aggregation behavior of IEEE 802.11n in a real-world, multi-hop, indoor mesh testbed. Furthermore, this thesis presents MPDU payload adaptation (MPA) to utilize A-MPDU subframes to increase the overall throughput under bad channel conditions. MPA adapts the size of MAC protocol data units to channel conditions, to increase the throughput and lower the delay in error-prone channels. The results suggest that under erroneous conditions throughput can be maximized by limiting the MPDU size.
As second major contribution, this thesis introduces Neighborhood-aware OPPortunistic networking on Smartphones (NOPPoS). NOPPoS creates an opportunistic, pocket-switched network using current generation, off-the-shelf mobile devices. As main novel feature, NOPPoS is highly responsive to node mobility due to periodic, low-energy scans of its environment, using Bluetooth Low Energy advertisements.
The last major contribution is the Neighborhood Document Sharing (NDS) protocol. NDS enables users to discover and retrieve arbitrary documents shared by other users in their proximity, i.e. in the communication range of their IEEE 802.11 interface. However, IEEE 802.11 connections are only used on-demand during file transfers and indexing of files in the proximity of the user. Simulations show that NDS interconnects over 90 \% of all devices in communication range.
Finally, NDS is extended by the content recommendation system User Preference-based Probability Spreading (UPPS), a graph-based approach. It integrates user-item scoring into a graph-based tag-aware item recommender system. UPPS utilizes novel formulas for affinity and similarity scoring, taking into account user-item preference in the mass diffusion of the recommender system. The presented results show that UPPS is a significant improvement to previous approaches.
| Identifer | oai:union.ndltd.org:DRESDEN/oai:qucosa:de:qucosa:72056 |
| Date | 04 September 2020 |
| Creators | Friedrich, Jan |
| Contributors | Lindemann, Christoph, Universität Leipzig |
| Source Sets | Hochschulschriftenserver (HSSS) der SLUB Dresden |
| Language | English |
| Detected Language | English |
| Type | info:eu-repo/semantics/acceptedVersion, doc-type:doctoralThesis, info:eu-repo/semantics/doctoralThesis, doc-type:Text |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0016 seconds