This article presents a family of computational storage drives (CSDs) and demonstrates their performance and power improvements due to in-storage processing (ISP) when running big data analytics applications. CSDs are an emerging class of solid state drives that are capable of running user code while minimizing data transfer time and energy. Applications that can benefit from in situ processing include distributed training, distributed inferencing, and databases. To achieve the full advantage of the proposed ISP architecture, we propose software solutions for workload balancing before and at runtime for training and inferencing applications. Other applications such as sharding-based databases can readily take advantage of our ISP structure without additional tooling. Experimental results on different capacity and form factors of CSDs show up to 3.1× speedup in processing while reducing the energy consumption and data transfer by up to 67% and 68%, respectively, compared to regular enterprise solid state drives.

The IPv6 routing protocol for low power and lossy networks (RPL) has some shortcomings, such as high packet loss rate and low network lifetime when used in Internet-of-Things (IoT) environments under heavy traffic. To overcome the RPL limitations, the current research tends to focus on a new paradigm of routing, referred to as Anycast Routing, in which a source node targets a set of destinations rather than a single one. In this article, we present a protocol that exploits the anycast perspective in the routing process since an important aspect of most IoT environments is to forward packets to a gateway, no matter which. Besides, we interconnect various instances of RPL to achieve better routing performance by offering the possibility of cooperation among various simultaneous instances of RPL within the network. Finally, to reach higher performance, we use a rank computation and parent selection mechanism that is different from those of the RPL. The evaluation results, which are obtained through simulation with the Cooja simulator, show that EM-RPL outperforms RPL in reducing the environmental footprint of the network, while it extends the network lifetime, decreases packet loss ratio, and better controls interpacket intervals and parent change overhead.

Computational storage devices enable in-storage processing of data in place. These devices contain 64-bit application processors and hardware accelerators that can help improving performance and saving power by reducing or eliminating data movement between host computers and storage units. This paper proposes a framework, named Stannis, for distributed in-storage training of deep neural networks on clusters of computational storage devices. This in-storage processing style of training ensures that private data never leaves the storage while fully controlling the public sharing of data. The Stannis framework distributes the workload based on the processing power of each worker by determining the proper batch size for each node. Stannis also ensures the availability of input data for all nodes to avoid rank stall while maximizing the utilization and overall processing speed. Experimental results show up to 2.7x speedup and 69% reduction in energy consumption with no significant loss in accuracy.

Distributed training is a novel approach to accelerating training of Deep Neural Networks (DNN), but common training libraries fall short of addressing the distributed nature of heterogeneous processors or interruption by other workloads on the shared processing nodes. This paper describes distributed training of DNN on computational storage devices (CSD), which are NAND flash-based, high-capacity data storage with internal processing engines. A CSD-based distributed architecture incorporates the advantages of federated learning in terms of performance scalability, resiliency, and data privacy by eliminating the unnecessary data movement between the storage device and the host processor. The paper also describes Stannis, a DNN training framework that improves on the shortcomings of existing distributed training frameworks by dynamically tuning the training hyperparameters in heterogeneous systems to maintain the maximum overall processing speed in term of processed images per second and energy efficiency. Experimental results on image classification training benchmarks show up to 3.1x improvement in performance and 2.45x reduction in energy consumption when using Stannis plus CSD compare to the generic systems.

The growing volume of data produced continuously in the Cloud and at the Edge poses significant challenges for large-scale AI applications to extract and learn useful information from the data in a timely and efficient way. The goal of this article is to explore the use of computational storage to address such challenges by distributed near-data processing. We describe Newport, a high-performance and energy-efficient computational storage developed for realizing the full potential of in-storage processing. To the best of our knowledge, Newport is the first commodity SSD that can be configured to run a server-like operating system, greatly minimizing the effort for creating and maintaining applications running inside the storage. We analyze the benefits of using Newport by running complex AI applications such as image similarity search and object tracking on a large visual dataset. The results demonstrate that data-intensive AI workloads can be efficiently parallelized and offloaded, even to a small set of Newport drives with significant performance gains and energy savings. In addition, we introduce a comprehensive taxonomy of existing computational storage solutions together with a realistic cost analysis for high-volume production, giving a good big picture of the economic feasibility of the computational storage technology.

Cloud applications are increasingly playing a crucial role in big data analytics. New use cases such as autonomous cars and edge computing call for novel approaches mixing heterogeneous computing and machine learning. These applications typically process petabyte-scale datasets, therefore, requiring low-power and scalable storage providing low-latency and high-throughput data access. While data centers have been focusing on migrating from legacy HDDs and SATA SSDs by deploying high-throughput and low-latency NVMe SSDs, the data bottlenecks appear as capacity scales. One approach to tackle this problem is to enable processing to happen within the storage device -in-storage processing (ISP)- eliminating the need to move the data. In this paper, we investigated the deployment of storage units with embedded low-power application processors along with FPGA-based reconfigurable hardware accelerators to address both performance and energy efficiency. To this purpose, we developed a high-capacity solid-state drive (SSD) named Catalina equipped with a quad-core ARM A53 processor running a Linux operating system along with a highly efficient FPGA accelerator for running applications in-place. We evaluated our proposed approach on a case study application for a similarity search library called Faiss.

Delay-sensitive video transmission over IEEE 802.11 wireless local area networks (WLANs) is analyzed in a cross-layer optimization framework. The effect of delay constraint on the quality of received packets is studied by analyzing “expired-time packet discard rate”. Three analytical models are examined and it is shown that M/M/1 model is quite an adequate model for analyzing delay-limited applications such as live video transmission over WLAN. The optimal MAC retry limit corresponding to the minimum “total packet loss rate” is derived by exploiting both mathematical analysis and NS-2 simulations. We have shown that there is an interaction between "packet overflow drop" and "expired-time packet discard" processes in the queue. Subsequently, by introducing the concept of virtual buffer size, we will obtain the optimal buffer size in order to avoid "packet overflow drop". We finally introduced a simple and yet effective real-time algorithm for retry-limit adaptation over IEEE 802.11 MAC in order to maintain a loss protection for delay-critical video traffic transmission, and showed that the average link-layer throughput can be improved by using our adaptive scheme.

We find a low-complicity and accurate model to solve the problem of optimizing MAC-layer transmission of real-time video over wireless local area networks (WLANs) using cross-layer techniques. The objective in this problem is to obtain the optimal MAC retry limit in order to minimize the total packet loss rate. First, the accuracy of Fluid and M/M/1/K analytical models is examined. Then we derive a closed-form expression for service time in WLAN MAC transmission, and will use this in mathematical formulation of our optimization problem based on M/G/1 model. Subsequently we introduce an approximate and simple formula for MAC-layer service time, which leads to the M/M/1 model. Compared with M/G/1, we particularly show that our M/M/1-based model provides a low-complexity and yet quite accurate means for analyzing MAC transmission process in WLAN. Using our M/M/1 model-based analysis, we derive closed-form formulas for the packet overflow drop rate and optimum retry-limit. These closed-form expressions can be effectively invoked for analyzing adaptive retry-limit algorithms. Simulation results (network simulator-2) will verify the accuracy of our analytical models.

An analytical model is employed to solve a cross-layer optimization problem in IEEE 802.11 wireless local area networks (WLANs). Closed-form expressions for the optimum retry limit, packet overflow drop rate and overall loss rate are derived using M/M/l queuing model, and subsequently an adaptive MAC retry limit scheme is studied. Furthermore simulation results (network simulator-2) will verify the accuracy of our analytical model.

Bit error rate (BER)-based constrained optimization of average spectral efficiency for adaptive Multi-level Quadrature Amplitude Modulation (MQAM) with service rate and power constraint is studied under composite Gamma-Lognormal fading channel condition. We obtain the corresponding optimal schemes for power and rate adaptation for continuous and discrete rate MQAM. The optimal power for continuous rate QAM is shown to be a combination of two functions; one of which with water-filling nature and the other proportional to log(1/BER), whereas for discrete rate QAM the solution is proportional to log(1/BER) over each rate region. We have also shown that transmission rate is a monotonically increasing function of BER.