Bibliographic record
Abstract
Scalable on-demand content delivery systems, designed to effectively handle increasing request rates, typically use service aggregation or content replication techniques. Service aggregation relies on one-to-many communication techniques, such as multicast, to efficiently deliver content from a single sender to multiple receivers. With replication, multiple geographically distributed replicas of the service or content share the load of processing client requests and enable delivery from a nearby server. Previous scalable protocols for downloading large, popular files from a single server include batching and cyclic multicast. Analytic lower bounds developed in this thesis show that neither of these protocols consistently yields performance close to optimal. New hybrid protocols are proposed that achieve within 20% of the optimal delay in homogeneous systems, as well as within 25% of the optimal maximum client delay in all heterogeneous scenarios considered. In systems utilizing both service aggregation and replication, well-designed policies determining which replica serves each request must balance the objectives of achieving high locality of service, and high efficiency of service aggregation. By comparing classes of policies, using both analysis and simulations, this thesis shows that there are significant performance advantages in using current system state information (rather than only proximities and average loads) and in deferring selection decisions when possible. Most of these performance gains can be achieved using only “local” (rather than global) request information. Finally, this thesis proposes adaptations of already proposed peer-assisted download techniques to support a streaming (rather than download) service, enabling playback to begin well before the entire media file is received. These protocols split each file into pieces, which can be downloaded from multiple sources, including other clients downloading the same file. Using simulations, a candidate protocol is presented and evaluated. The protocol includes both a piece selection technique that effectively mediates the conflict between achieving high piece diversity and the in-order requirements of media file playback, as well as a simple on-line rule for deciding when playback can safely commence.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.000 | 0.005 |
| Open science | 0.006 | 0.004 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".