RJRayana H. Jaafar

Rayana H. Jaafar

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Rayana H. Jaafar has a total of 20 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Federated Learning with Incrementally Aggregated Gradients

    by Aritra Mitra, Rayana Jaafar, George J. Pappas, Hamed Hassani

    We consider the standard federated learning (FL) framework where a set of clients coordinate with a central server to train a statistical model. In a single-machine centralized setting, it is well-known that for smooth and strongly convex finite-sum optimization problems, one can design algorithms that guarantee exact linear convergence to the global minimum without computing full (batch) gradients at every iteration. Despite its popularity, an analog of the above result does not exist in FL. Motivated by this gap, we consider a setting where the local loss function of each client can be expressed as a finite sum of smooth component functions. For this setting, we propose a novel computationally-efficient FL algorithm called FedTrack that rests on two key ideas: (i) using the most recently communicated versions of the clients’ gradients in the local update rule, and (ii) incrementally aggregating gradients of the component functions of each client. While the first idea serves to overcome the effect of heterogeneity across the clients’ local loss functions, the second helps to significantly reduce the overall number of gradient computations. For both strongly convex and non-convex local loss functions, we prove that the convergence guarantees of FedTrack match their centralized counterparts (up to constants). In particular, for the strongly convex setting, we show that FedTrack guarantees exact linear convergence to the global minimum deterministically.

  • Which journal ranking list? A case study in business and economics

    by Rayana Jaafar, Vijay Pereira, Samer S. Saab, Abdul-Nasser El-Kassar

    Purpose: With over 3,000 academic journals in the fields of Business and Economics, most academics face a hard time selecting an adequate journal to submit their work to. In today's demanding academic environment and with the presence of different journal ranking lists (JRLs), the selection becomes more difficult when considering employment, promotion and funding. The purpose of this paper is to explore key differences among multiple JRLs pertinent to the latter common objectives. An extensive analysis is conducted to compare the content of journals in the Australian Business Deans Council (ABDC) Journal Quality list, Scopus and Web of Science (WoS) in the fields of Business and Economics. Then, a case of a university with medium research output is considered where scholarly performance evaluation is based on the ABDC Journal Quality List.Design/methodology/approach: After ranking journals in the fields of Business and Economics based on SCImago Journal Rank (SJR) indicator, JCR's Journal Impact Factor (JIF) and JCR's Eigenfactor (EF), a methodology is proposed to categorize journals in the three JRLs into the same categorization adopted by ABDC. The latter establishes a way to compare the four JRLs under consideration and serves as a basis to compare and analyze the content of journals in the ABDC Journal Quality list, Scopus and WoS. As a proxy impact metric, a normalized citation count is associated with each article based on Google Scholar. The publications of the considered university are then evaluated from the perspective of the four JRLs in terms of citation-based impact and quality while considering the exposure to popular world university ranking tables.Findings: For journals classified under fourth tier by ABDC, over 53 and 59% are not indexed by Scopus and WoS, respectively. In this case study, over 42% of the publications appear in journals that are not listed in JCR despite the fact that over 94% of them are listed by the SJR list. Generally, publications that appear in journals listed by JCR achieve, on a yearly average, significantly higher citation rates when compared to those that appear in journals listed in ABDC and SJR Lists.Originality/value: A four-tier mapping is proposed for consistent comparison among JRLs. Normalized citation count associated with each article based on Google Scholar is employed for evaluation. The findings provide recommendations for scholars, administrators and global universities, including Euro-Med Universities, on which JRL can be more influential for both faculty development and positioning of the university.


  • A proportional-derivative-double derivative controller for robot manipulators

    by Samer S. Saab, Rayana H. Jaafar

    Feeding a controller with acceleration signal can improve system robustness and relative stability. However, neither measurements of acceleration signals nor velocity signals are available in most industrial robot manipulators. Inevitable estimation errors may lead to chattering phenomenon that can excite unmodeled high frequency plant dynamics yielding either severe vibrations in the arm or instability of the control system. This paper proposes a stochastic discrete-time multivariable proportional-derivative-double derivative (PDD) controller with the objective of possessing superior tracking capability in presence of unmodeled joint friction forces while only using joint position measurements. The proposed approach models the estimation errors of velocity and acceleration stochastically. Subsequently, the time-varying PDD gains are derived based on minimising the mean-square state error per-discrete-time instant. We show that the mean-square velocity error is inversely proportional to the square of the sample rate. We demonstrate the tracking superiority of the proposed control experimentally on a four-degree-of-freedom robot manipulator.


  • Massive Open Online Labs (MOOLs): An Innovative Solution to Achieving SDGs in the Global South

    by Fariba Moghaddam, Nomaou Dan Lamso, Hassane Alzouma Mayaki, Hassan Ali Barkad, Hamadou Saliah Hassane, Denis Gillet, Aldo Vaccari, Thomas Moniquet, Christophe Salzmann, Yves Piguet, et al.

    In engineering education, laboratories represent an important academic resource as they provide practical training in addition to the fundamental theories. However, the acquisition of new machinery and the maintenance of the equipment imply a large investment that only a limited number of universities can afford. This paper represents innovative online education activities through a collaborative widespread network with the global south countries, deploying remote laboratories in electrical, mechanical and control engineering at a large scale within MOOC infrastructures.


  • Sliding Mode Control with Dirty Derivatives Filter for Rigid Robot Manipulators

    by Rayana H. Jaafar, Samer S. Saab

    Although Sliding Mode Control (SMC) is shown to be a powerful method for the control of rigid robotic manipulators, it comes with disadvantages. One disadvantage is related to the fact that the terminal sliding variable vector has a nonlinear term of the velocity error. Consequently, inaccurate velocity estimation may lead to chattering phenomenon that can excite unmodeled high-frequency plant dynamics yielding to either severe vibrations in the arm or instability of the control system. To better overcome such a drawback, this paper proposes the integration of dirty derivatives filter with SMC instead of using "direct" velocity estimation methods. We justify the effectiveness of our proposed approach numerically and experimentally on a four-degree-of-freedom (DOF) robot manipulator while considering a classical SMC and two adaptive SMCs.


  • A Neural Network Approach for Indoor Fingerprinting-Based Localization

    by Rayana H. Jaafar, Samer S. Saab

    Indoor Positioning System (IPS) has been continuously growing in importance, and the ability of accurate localization has become a key issue. The received signal strength (RSS) fingerprinting based methodology is widely considered, and many related localization techniques have been studied. One of the most challenging tasks in fingerprint-based IPS is accurate localization in an open area; e.g., with no walls, where fingerprints may not be as scattered as desired. The latter can be moderated by using signals of opportunities, and since most user equipment (UE) comprise different wireless access technologies, fingerprinting using heterogeneous wireless networks becomes viable. On the other hand, RSS fingerprinting based localization becomes rather complex when processed mathematically such as radio frequency propagation models. This paper proposes the adoption of a neural network for a standalone RSS fingerprinting-based IPS. We evaluate the performance of our proposed method in realistic environment using WLAN and/or 4G networks. Our experimental results show that sub-meter accuracy can be achieved whenever the RSS fingerprints are composed of both networks.


  • Open Access

    A Multiple‐Access Time and Frequency Spectrum‐Spreading Modulation

    by Samer S Saab, Joe Khalife, Rayana H Jaafar

    In this paper a novel modulation scheme called Carrier Interleaved Multiple Access (CIMA) is proposed. CIMA provides an alternative for multiple‐access modulation accommodating resistance to noise and channel interference. The approach is based on polar signaling modulated with an FM‐like composite sinusoidal function. The user assigned frequency deviation and modulation index are strictly related and unique. The latter parameters are generated using a nontraditional pseudorandom noise generator (PRNG). This PRNG provides CIMA with low interference capability between cochannels and adjacent channels. CIMA can be considered for a single‐user or multiple‐access technique. Selected CIMA characteristics are presented. In order to numerically illustrate the effectiveness of the proposed modulation scheme, the performance of CIMA is compared with the conventional direct‐sequence spread spectrum binary phase‐shift keying (DSSS‐BPSK) modulation.