Bibliographic record
Abstract
Crowdsensing, the use of everyday devices to collect and share data is paving the way for cost-efficient real-time data collection. Every day, information can be quickly sensed and shared publicly using smartphones. Beyond smartphones, modern vehicles have also shown great promise for crowdsensing. In contrast to mobile crowdsensing, vehicles are ideal platforms to collect, store, compute, and share large amounts of sensor data. Vehicles have greater mobility and cover wider sensing area. The mobility patterns of vehicles are predictable due to the prevalence of navigation systems. Most importantly, the abundance of on-board resources and lack of power constraints makes it possible to support complex and long-lasting sensing tasks. The idea behind vehicular crowdsensing is to leverage vehicles as mobile sensors and computing resources. Vehicles are recruited as sensing participants for large-scale crowdsensing tasks such as urban sensing or traffic condition monitoring. However, existing works often assume that sensing tasks are common tasks where sensing results are shared with the public. Instead of public information, we believe the benefit of the crowdsensing paradigm should be available for personal use. In this, we focus on small-scale sensing tasks, tasks that dynamically change over time and are unlikely to be shared. Thus, sensing information is collected on demand rather than continuously and at all times. We refer to this as personalized vehicular crowdsensing. Furthermore, most works assume the routing of vehicular participants cannot be changed. This further reduces a system's ability to fulfill dynamic sensing needs. Thus, in this thesis, we further explore the possibilities of improving crowdsensing performance by vehicle route planning. To achieve either public vehicular crowdsensing or personalized vehicular crowdsensing, we must resolve the problem of vehicular participant selection. We first propose two solutions based on push and pull for vehicular crowdsensing. We aim to maximize sensing coverage, improve load balance, error tolerance, and minimize costs. Next, we improve the sensing performance by allowing vehicles to reroute. Instead of greedily generating routes to maximize overall sensing coverage, we leverage a two-step global and local planning algorithm. Our global algorithm attempts to plan a vehicle's route based on the difference between the distribution of vehicular location and the distribution of task location. Our goal is to send vehicles to areas with a high number of tasks but having few vehicles to service them. The global algorithm does not determine which tasks it should service; this operation is handled by the local algorithm which decides how to optimally leverage a vehicle's sensing ability for a small area. Through the use of SUMO simulation and TAPAS Cologne Large Scale Mobility Dataset, we show that our proposed approaches deliver significant performance improvements compared to traditional approaches
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.003 | 0.006 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.012 | 0.006 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".