Multiple Round Trips Research Articles

Performance Evaluation of TCP BBRv3 in Networks with Multiple Round Trip Times

The Transmission Control Protocol (TCP) serves as a cornerstone mechanism for implementing Congestion Control (CC) across the Internet. Designing a solution that provides high bandwidth utilization and mitigates the phenomenon of bufferbloat across a spectrum of diverse scenarios poses a considerable challenge. The introduction of Bottleneck Bandwidth and Round Trip propagation time (BBR) in 2016 marked a significant shift in congestion control methodology. Its improved performance and adaptability contributed to the initial acclaim and widespread interest that it received.. Unlike most currently used CCs, it operates around Kleinrock’s optimal point, thus offering high throughput even in lossy networks while preventing buffer saturation. Unfortunately, it quickly became evident that BBR was unable to fairly share bandwidth with flows characterized by different path delays, as well as loss-based CCs. In response, Google recently introduced a third iteration to address these shortcomings. This study explores the performance of BBRv3 across a wide range of scenarios, thereby considering different buffer sizes and paths with varying Round Trip Times (RTTs), and it evaluates its superiority over its predecessors. Through extensive simulations, this work assesses whether BBRv3 can finally play fair with other bandwidth contenders, which is a critical consideration given the widespread deployment of BBR. The framework is publicly available to facilitate additional validation and ensure the reproducibility of the study’s findings. The results indicate that while BBRv3 demonstrates enhanced fairness towards loss-based CC algorithms, it struggles when competing against other BBR flows, especially in multi-RTT networks, thus falling short even when compared to the initial version.

Fast Commitment for Geo-Distributed Transactions via Decentralized Co-Coordinators

In a geo-distributed database, data shards and their respective replicas are deployed in distinct datacenters across multiple regions, enabling regional-level disaster recovery and the ability to serve global users locally. However, transaction processing in geo-distributed databases requires multiple cross-region communications, especially during the commit phase, which can significantly impact system performance. To optimize the performance of geo-distributed transactions, we propose Decentralized Two-phase Commit (D2PC), a new transaction commit protocol aiming to minimize the negative impact of cross-region communication. In D2PC, we employ multiple co-coordinators that perform commit coordination in parallel. Each co-coordinator is responsible for collecting 2PC votes and making a PreCommit decision in its local region. This approach allows for the concurrent invocation of multiple cross-region network round trips, and each region can end its concurrency control locally before replication is complete, thus significantly reducing the chances of blocking and enhancing system concurrency. Moreover, we propose the bypass leader replication reply method, leveraging decentralized co-coordinators to bypass the leader for message transmission, thereby reducing the commit latency. Experimental results have demonstrated that D2PC can reduce commit latency by 43% and improve throughput by up to 2.43 × compared to the geo-distributed transaction processing methods based on 2PC.

Multi-bounce self-mixing in terahertz metasurface external-cavity lasers

The effects of optical feedback on a terahertz (THz) quantum-cascade metasurface vertical-external-cavity surface-emitting laser (QC-VECSEL) are investigated via self-mixing. A single-mode 2.80 THz QC-VECSEL operating in continuous-wave is subjected to various optical feedback conditions (i.e., feedback strength, round-trip time, and angular misalignment) while variations in its terminal voltage associated with self-mixing are monitored. Due to its large radiating aperture and near-Gaussian beam shape, we find that the QC-VECSEL is strongly susceptible to optical feedback, which is robust against misalignment of external optics. This, in addition to the use of a high-reflectance flat output coupler, results in high feedback levels associated with multiple round-trips within the external cavity-a phenomenon not typically observed for ridge-waveguide QC-lasers. Thus, a new theoretical model is established to describe self-mixing in the QC-VECSEL. The stability of the device under variable optical feedback conditions is also studied. Any mechanical instabilities of the external cavity (such as vibrations of the output coupler), are enhanced due to feedback and result in low-frequency oscillations of the terminal voltage. The work reveals how the self-mixing response differs for the QC-VECSEL architecture, informs other systems in which optical feedback is unavoidable, and paves the way for QC-VECSEL self-mixing applications.

A High-performance RDMA-oriented Learned Key-value Store for Disaggregated Memory Systems

Disaggregated memory systems separate monolithic servers into different components, including compute and memory nodes, to enjoy the benefits of high resource utilization, flexible hardware scalability, and efficient data sharing. By exploiting the high-performance RDMA (Remote Direct Memory Access), the compute nodes directly access the remote memory pool without involving remote CPUs. Hence, the ordered key-value (KV) stores (e.g., B-trees and learned indexes) keep all data sorted to provide range query services via the high-performance network. However, existing ordered KVs fail to work well on the disaggregated memory systems, due to either consuming multiple network roundtrips to search the remote data or heavily relying on the memory nodes equipped with insufficient computing resources to process data modifications. In this article, we propose a scalable RDMA-oriented KV store with learned indexes, called ROLEX, to coalesce the ordered KV store in the disaggregated systems for efficient data storage and retrieval. ROLEX leverages a retraining-decoupled learned index scheme to dissociate the model retraining from data modification operations via adding a bias and some data movement constraints to learned models. Based on the operation decoupling, data modifications are directly executed in compute nodes via one-sided RDMA verbs with high scalability. The model retraining is hence removed from the critical path of data modification and asynchronously executed in memory nodes by using dedicated computing resources. ROLEX efficiently alleviates the fragmentation and garbage collection issues, due to allocating and reclaiming space via fixed-size leaves that are accessed via the atomic-size leaf numbers. Our experimental results on YCSB and real-world workloads demonstrate that ROLEX achieves competitive performance on the static workloads, as well as significantly improving the performance on dynamic workloads by up to 2.2× over state-of-the-art schemes on the disaggregated memory systems. We have released the open-source codes for public use in GitHub.

Transport of Multispecies Ion Crystals through a Junction in a Radio-Frequency Paul Trap.

We report on the first demonstration of transport of a multispecies ion crystal through a junction in a rf Paul trap. The trap is a two-dimensional surface-electrode trap with an X junction and segmented control electrodes to which time-varying voltages are applied to control the shape and position of potential wells above the trap surface. We transport either a single ^{171}Yb^{+} ion or a crystal composed of a ^{138}Ba^{+} ion cotrapped with the ^{171}Yb^{+} ion to any port of the junction. We characterize the motional excitation by performing multiple round-trips through the junction and back to the initial well position without cooling. The final excitation is then measured using sideband asymmetry. For a single ^{171}Yb^{+} ion, transport with a 4 m/s average speed induces between 0.013±0.001 and 0.014±0.001 quanta of excitation per round-trip, depending on the exit port. For a Ba-Yb crystal, transport at the same speed induces between 0.013±0.001 and 0.030±0.002 quanta per round-trip of excitation to the in-phase axial mode. Excitation in the out-of-phase axial mode ranges from 0.005±0.001 to 0.021±0.001 quanta per round-trip.

Dynamics and feedback sensitivity of vertical-cavity surface-emitting lasers with arbitrary optical feedback intensity

Vertical cavity surface emitting lasers (VCSELs) are the key laser sources in the high-performance computers (HPCs) and data centers (DCs). The explosive growth of data traffic in the HPCs and DCs requires increased bandwidths of VCSELs. In this paper, we deduce the Time-Delay (T-D) model considering the multiple round trips of light in VCSELs with an optical feedback cavity to enhance the modulation bandwidth. The T-D model is suitable for the VCSEL with an arbitrary optical feedback intensity. The influences of the round trip time m of light in the optical feedback cavity on the intensity-modulation responses with strong and weak optical feedback are studied. The feedback coefficients considering the multiple round trips of light in the optical feedback cavity and the dispersions of the distributed Bragg reflector are deduced, and are suitable for the arbitrary optical feedback intensity. The feedback coefficients are influenced by the value of m and the optical feedback intensity, and decrease as the value of m under arbitrary optical feedback intensity increases. The feedback coefficients under strong optical feedback are larger than that under weak optical feedback with the same value of m. The theories are useful for the high-speed VCSELs with optical feedback.

On the Evaluation of a Drone-Based Delivery System on a Mixed Euclidean-Manhattan Grid

In this work, we investigate the use of drones in a delivery scenario formed by two contiguous areas. In one area the drones can freely fly on straight lines between any two locations (Euclidean metric), while in the other one the drones must follow the open space above the roads (Manhattan metric). We model this delivery scenario as a Euclidean-Manhattan-Grid (EM-grid). Given a set of customers to be served in an EM-grid, the objective is to find the distribution point (DP) for the drone that minimizes the overall traveled distance, considering that the drone has to do multiple round trips to/from the DP. In our view, the DP is optimized with respect to the set of customers and its computation must be light because it needs to be recomputed every time the set of customers varies. Accordingly, we define the Single Distribution Point Problem (SDPP) and devise sub-optimal time-efficient algorithms for solving it. We numerically compare the cost of our sub-optimal solutions with that of an optimal solution computed with a brute-force approach. Finally, using the BlueSky open air simulator, we compare the cost of our best solution with the cost of a solution that serves the costumers from a fixed DP, like the location of a delivery company’s depot. The fixed DP can perform very poorly for some customer instances, while our solution is highly adaptive and reduces the time and the distance covered by the drone.

Self-Pulsations in Terahertz Quantum Cascade Lasers under Strong Optical Feedback: The Effect of Multiple Reflections in the External Cavity.

We have recently reported the self-pulsation phenomenon under strong optical feedback in terahertz (THz) quantum cascade lasers (QCLs). One important issue, however, we left open: the effect of multiple round trips in the external cavity on the laser response to feedback. Our current analysis also casts additional light on the phenomenon of self-pulsations. Using only one external cavity round trip (ECRT) in the model has been the common approach following the seminal paper by Lang-Kobayashi in 1980. However, the conditions under which the Lang-Kobayashi model, in its original single-ECRT formulation, is applicable has been rarely explored. In this work, we investigate the self-pulsation phenomenon under multiple ECRTs. We found that the self-pulsation waveform changes when considering more than one ECRT. This we attribute to the combined effect of the extended external cavity length and the frequency modulation of the pulsation frequency by the optical feedback. Our findings add to the understanding of the optical feedback dynamics under multiple ECRTs and provide a pathway for selecting the appropriate numerical model to study the optical feedback dynamics in THz QCLs and semiconductor lasers in general.

Non-classical mechanical states guided in a phononic waveguide

The ability to create, manipulate and detect non-classical states of light has been key for many recent achievements in quantum physics and for developing quantum technologies. Achieving the same level of control over phonons, the quanta of vibrations, could have a similar impact, in particular on the fields of quantum sensing and quantum information processing. Here we present a crucial step towards this level of control and realize a single-mode waveguide for individual phonons in a suspended silicon microstructure. We use a cavity–waveguide architecture, where the cavity is used as a source and detector for the mechanical excitations while the waveguide has a free-standing end to reflect the phonons. This enables us to observe multiple round trips of phonons between the source and the reflector. The long mechanical lifetime of almost 100 μs demonstrates the possibility of nearly lossless transmission of single phonons over, in principle, tens of centimetres. Our experiment demonstrates full on-chip control over travelling single phonons strongly confined in the directions transverse to the propagation axis, potentially enabling a time-encoded multimode quantum memory at telecommunications wavelength and advanced quantum acoustics experiments. Non-classical vibrations are generated and transmitted along a mechanical waveguide, providing a platform for distributing quantum information and realizing hybrid quantum devices using phonons in a solid-state system.

Radiofrequency modulator for marine lidar radar systems featuring compact and agile extra-cavity architecture using a polarimetric effect.

This paper proposes a new, to the best of our knowledge, modulator architecture for a microwave-modulated lidar for marine applications. The principle is based on the use of an infrared picosecond laser source, coupled to an external cavity, ensuring wavelength conversion in the visible range as well as radiofrequency modulation. Wavelength conversion is performed by a nonlinear crystal associated with adapted mirrors, while multiple round trips and polarization control in the cavity ensure microwave modulation. This paper presents both modeling of the emitted signal and the practical realization of this device.

Seasonal Patterns in Daily Flight Distance and Space Use by Great Egrets (Ardea alba)

In an effort to quantify the value of wetland habitats, GPS technology was used to document the movement patterns of 16 Great Egrets (Ardea alba) in North America. Patterns in daily flight distances and utilization distributions (UD; estimates of area occupied on the ground) were documented throughout the annual cycle. Maximum Daily Displacement (MDD), the farthest distance occupied by a bird from colony/roost (central place) in 24 hr was greatest (4.3 ± 0.1 km) during breeding season and lowest (3.3 ± 0.1) during post-breeding season. As birds visited multiple foraging sites and made multiple round trips to central places, this Total Daily Distance (TDD) was also measured. It increased from a mean of about 12 km during the incubation phase to about 35 km at the time of fledging. Average TDD was greatest during breeding season (14.2 ± 0.3 km) and lowest during winter (11.0 ± 0.2 km). The utilization distribution increased from 128 (± 21.3) ha during breeding season to 179 (±32.6) ha during winter. Birds that foraged at tidal sites used 183.3 (± 22.2) ha, twice as much area as birds that foraged strictly in freshwater sites (89.6 ± 21.3 ha).

An Improved Congestion Control Algorithm based on SCPS-TP Protocol

mainly in Vegas agreement by predicting the real-time changes of RTT is to adjust the congestion window, is likely to lead to network to link data congestion has not taken place, but because in the process of reverse echo is sure that the information congestion resulting in RTT value, detection mechanism will default to forward congestion and reduce the congestion window. In order to avoid this situation, RTT acquisition is modified to solve the problem that the forward congestion window may be reduced due to the reverse link congestion, so that the congestion window can be adjusted more accurately, and the network prediction mechanism is added to carry out joint adjustment scheme for the congestion window. Multiple round trip delays are stored at the sending end, and the network information obtained by the original sendRTT is further classified and analyzed through the analysis and processing of the delay by the algorithm, so as to obtain the status of the link more accurately and make accurate adjustments to the congestion window.

Variation in movement patterns of mule deer: have we oversimplified migration?

BackgroundConservation and management of migratory animals has gained attention in recent years, but the majority of research has focused on stereotypical ‘migrant’ and ‘resident’ behaviors, often failing to incorporate any atypical behaviors or characterize migratory behaviors beyond distance and timing of the migration. With migration threatened by anthropogenic development and climate change, it is crucial that we understand the full range of migratory behaviors. Our objective was to demonstrate and characterize the variation in migration strategies, including typical and atypical migratory behaviors for mule deer (Odocoileus hemionus) in Utah, USA.MethodsBecause calculation of common metrics such as distance, timing, and use of stopovers during migration did not adequately describe the variation we observed in migratory behavior for this species—particularly when animals visited multiple (> 3) ranges for extended lengths of time—we developed additional methods and categories to describe observed variation in migratory behavior. We first categorized trajectories based on the number of discrete, separate ranges and range shifts between them. Then, we further characterized the variation in migration strategies by examining the timing, duration, and distance traveled within each of the categories. We also examined if and how frequently individual deer switched among categories from year to year.ResultsWe classified 1218 movement trajectories from 722 adult female mule deer, and found that 54.4% were dual-range migrants, who made one round-trip to one distinct range. Multi-range migrants (23.6%) made one round-trip during which they stayed at multiple discrete ranges. Commuters (1.0%) traveled to the same range multiple times, and poly migrants (1.5%) made multiple round-trips to different ranges. Gradual movers (2.5%) did not show a discrete range shift but moved gradually between ranges, whereas residents (12.6%) never left their home ranges, and dispersers (4.4%) left but never returned. Of the deer that we monitored for multiple years, 51.2% switched among categories.ConclusionWe conclude that the substantial number of atypical migratory strategies, as well as the number of deer that switched categories, underlines the importance of studying these less-stereotyped behaviors that may be exhibited by large proportions of populations. Acknowledging and investigating the full complexity and diversity in migratory strategies might uncover unknowns with respect to underlying factors and drivers of migration, and can help shape effective conservation strategies.

XStore : Fast RDMA-Based Ordered Key-Value Store Using Remote Learned Cache

RDMA ( Remote Direct Memory Access ) has gained considerable interests in network-attached in-memory key-value stores. However, traversing the remote tree-based index in ordered key-value stores with RDMA becomes a critical obstacle, causing an order-of-magnitude slowdown and limited scalability due to multiple round trips. Using index cache with conventional wisdom—caching partial data and traversing them locally—usually leads to limited effect because of unavoidable capacity misses, massive random accesses, and costly cache invalidations. We argue that the machine learning (ML) model is a perfect cache structure for the tree-based index, termed learned cache . Based on it, we design and implement XStore , an RDMA-based ordered key-value store with a new hybrid architecture that retains a tree-based index at the server to perform dynamic workloads (e.g., inserts) and leverages a learned cache at the client to perform static workloads (e.g., gets and scans). The key idea is to decouple ML model retraining from index updating by maintaining a layer of indirection from logical to actual positions of key-value pairs. It allows a stale learned cache to continue predicting a correct position for a lookup key. XStore ensures correctness using a validation mechanism with a fallback path and further uses speculative execution to minimize the cost of cache misses. Evaluations with YCSB benchmarks and production workloads show that a single XStore server can achieve over 80 million read-only requests per second. This number outperforms state-of-the-art RDMA-based ordered key-value stores (namely, DrTM-Tree, Cell, and eRPC+Masstree) by up to 5.9× (from 3.7×). For workloads with inserts, XStore still provides up to 3.5× (from 2.7×) throughput speedup, achieving 53M reqs/s. The learned cache can also reduce client-side memory usage and further provides an efficient memory-performance tradeoff, e.g., saving 99% memory at the cost of 20% peak throughput.

Beyond MPI

Networkswith Remote DirectMemoryAccess (RDMA) support are becoming increasingly common. RDMA, however, offers a limited programming interface to remote memory that consists of read, write and atomic operations. With RDMA alone, completing the most basic operations on remote data structures often requires multiple round-trips over the network. Data-intensive systems strongly desire higher-level communication abstractions that supportmore complex interaction patterns. A natural candidate to consider is MPI, the de facto standard for developing high-performance applications in the HPC community. This paper critically evaluates the communication primitives of MPI and shows that using MPI in the context of a data processing system comes with its own set of insurmountable challenges. Based on this analysis, we propose a new communication abstraction named RDMO, or Remote DirectMemory Operation, that dispatches a short sequence of reads, writes and atomic operations to remote memory and executes them in a single round-trip.

Quantum photonics with active feedback loops

We develop a unified theoretical framework for the efficient description of multiphoton states generated and propagating in loop-based optical networks which contain nonlinear elements. These active optical components are modeled as nonlinear media, resembling a two-mode squeezer. First, such nonlinear components can be seeded to manipulate quantum states of light, as such enabling photon addition protocols. And, second, they can function as an amplifying medium for quantum light. To prove the practical importance of our approach, the impact of multiple round trips is analyzed for states propagating in experimentally relevant loop configurations of networks, such as time-multiplexed driven quantum walks and iterative photon-number state generation protocols. Our method not only enables us to model such complex systems but also allows us to propose alternative setups that overcome previous limitations. To characterize the systems under study, we provide exact expressions for fidelities with target states, success probabilities of heralding-type measurements, and correlations between optical modes, including many realistic imperfections. Moreover, we provide an easily implementable numerical approach by devising a vector-type representation of photonic states, measurement operators, and passive and active processes.

Instabilities in a dissipative soliton-similariton laser using a scalar iterative map.

Numerical simulations of a dissipative soliton-similariton laser are shown to reproduce a range of instabilities seen in recent experiments. The model uses a scalar nonlinear Schrödinger equation map, and regions of stability and instability are readily identified as a function of gain and saturable absorber parameters. Studying evolution over multiple round trips reveals spectral instabilities linked with soliton molecule internal motion, soliton explosions, chaos, and intermittence. For the case of soliton molecules, the relative phase variation in the spectrum is shown to be due to differences in nonlinear phase evolution between the molecule components over multiple round trips.

Familiarity Breeds Short-Termism

Investors exhibit a robust and systematic pattern of shortening their holding period in a stock on which they execute multiple round trip trades. On average, the holding period shortens by 11% with each additional round trip. I show this tendency to be short-termed is associated with reinforcement learning. Investors are more likely to shorten the holding period after a round trip where they could have realized a better return had they sold earlier. Investors become short-termed as they become more familiar with trading a stock.

A hybrid heuristic optimization of maintenance routing and scheduling for offshore wind farms

As the operation and maintenance (O&M) costs constitute a substantial portion of the overall life-cycle cost of offshore wind farms, routing, and scheduling of maintenance are very important for cost reduction. With the multi-type of vessels, multi-period, multi-base of O&M, multi-wind farm and uncertain weather conditions, the optimization of O&M cost is more challenging. In this article, a hybrid heuristic optimization of maintenance routing and scheduling for offshore wind farms is proposed. First, with the maintenance service protocol, mixed particle swarm optimization (MPSO) is applied to seek a desired mapping relation between vessels and wind farms. Utilizing the formalized rules, an optimal vessel allocation scheme is explored in the large solution space by individual crossover, swarm crossover and mutation. Then, with the scheme of vessel allocation, a discrete wolf pack search (DWPS) is introduced to optimize the maintenance route under all constraints. As the evaluation standard of MPSO, the purpose of DWPS is to search the solution space with depth and breadth balanced and find the optimal and open maintenance route with multiple round trips to bases that minimize O&M costs, including travel, technician and penalty costs. Finally, computational experiments and analysis are carried out. The results provide both the optimized cost and detailed arrangements, which can be directly used in the maintenance schedule.

Modelling and characterisation of the noise characteristics of the vertical cavity surface-emitting lasers subject to slow light feedback

This paper introduces the modelling and characterisation of the noise properties of the vertical cavity surface-emitting laser (VCSEL) coupled in lateral direction with a passive cavity. This design of VCSEL with this transverse coupled cavity (TCC) is proposed for high-speed photonics. We introduce comprehensive simulations on the influence of the induced lateral slow light feedback on the relative intensity noise (RIN) and carrier-to-noise ratio (CNR). The proposed model incorporates multiple round trips of slow light in the TCC as time delay light in the rate equations of the VCSEL. The obtained results are compared with those of the conventional VCSEL. We show that the noise performance of the TCC-VCSEL is optimised when the VCSEL exhibits stable continuous wave (CW) operation under strong slow light feedback and the TCC length is smaller than 8 $$\mu $$ m and between 11 and 13 $$\mu $$ m. When strong slow light induces unstable regular and / or irregular oscillations, RIN is enhanced and CNR is lowered.