Loading...
Search for:
samandi--farid
0.119 seconds
Improving the Efficiency of On-chip 3D Stacked DRAM in Server Processors
, M.Sc. Thesis Sharif University of Technology ; Sarbazi-Azad, Hamid (Supervisor) ; Lotfi Kamran, Pejman (Co-Advisor)
Abstract
Big-data server workloads have vast datasets, and hence, frequently access off-chip memory for data. Consequently, server workloads lose significant performance potential due to off-chip latency and bandwidth walls. Recent research advocates using 3D stacked DRAM to break the walls. As 3D stacked DRAM cannot accommodate the whole datasets of server workloads, most proposals use 3D DRAM as a large cache. Unfortunately, a large DRAM cache imposes latency overhead due to (1) the need for tag lookup and (2) inefficient utilization of on-chip and off-chip bandwidth, and as a result, lowers the benefits of 3D stacked DRAM. Moreover, storing the tags of a multi-gigabyte DRAM cache requires changes...
Throughput Analysis in Multi Hop Cellular Networks
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
OFDMA multi-hop cellular network (MCN) is the topic of a wide range of research today. In this thesis, we analyse this type of networks, regarding shadowing status, mobility patterns and different service requirements. In the multi-hop (MH) scenario, six fixed relays are set up in each cell, which split the cell into seven subcells. For mobility modeling, each subcell is split into several sub-regions. Mobile nodes are able to maintain at each sub-region with a random time distribution and migrate from it to one of its neighboring sub-regions. We also consider three types of services requiring specific number of subcarriers with specific BERs. The set of subcarriers is distributed among...
Throughput Analysis of a Cognitive Multi-hop Wireless Network Overlaid on a Cellular Network
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
In this thesis, we evaluate the maximum stable throughput of a cognitive multi-hop IEEE 802.11-based WLAN overlaid on a cellular network. In the considered scenario, the secondary users operate in the downlink or uplink frequency band of primary network and transmit their data in free primary channels using OFDM technique. The activity of primary nodes is modeled independently by ON-OFF alternating states. To model the scenario, each secondary user is modeled with an open queueing network. The queueing network can model the transmission of data as well as the effect of primary users. By writing the traffic equations of the queueing network and applying the stability conditions, we are able...
Analysis of the Effect of Frame Aggregation in IEEE 802.11 Standard
, M.Sc. Thesis Sharif University of Technology ; Ashtiyani, Farid (Supervisor)
Abstract
In this thesis the effect of frame aggregation mechanism in maximizing throughput in Wireless Local Area Networks (WLANs) have been investigated. Frame aggregation has become a part of the IEEE 802.11 standard as the amendment IEEE 802.11n. First, we have a quick review on access control layer in IEEE 802.11 standard then illustrate the necessity of frame aggregation. Afterwards, we propose an analytical model for the analysis of an IEEE 802.11n network comprised of an Access Point (AP) and several conventional nodes, all in the coverage area of each other. This analytical model focuses on downlink throughput; therefore only AP uses frame aggregation and the other nodes use basic IEEE 802.11...
Network Coding-based Packet Relaying in Wireless Networks
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
In this thesis, we propose two new network coding-based packet relaying schemes suitable for multiple unicast scenarios in a random access network. In such networks the channels may have memory dependent upon the random access scheme. Thus, the optimal scheme for memoryless channels reported recently in the literature is not necessarily optimal in random access networks. In our proposed schemes, we convert a multiple unicast scenario to a combination of multiple anycast and a few multicast scenarios. By such an approach, we exploit the channels between relay node and the other nodes, instead of using only one channel at each packet transmission, for as many as possible. By simple analytical...
Content Distribution Service Analysis in Vehicular Ad-Hoc Networks
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
Car accident damages take about 7 percent of GPD of Iran every year. As a result of car accidents, the economy of the country is damaging irreparably. The amount of road casualties equals casualties in a complete war. Considering the aforementioned conditions, the researchers are thinking about new technologies to reduce the cost of the accidents. So the use of Vehicular ad-hoc Networks (VANET) is inevitable in the near future. Content distribution service is one of the emerging services of VANETs. This service is used in both safety and comfort related applications. Content distribution describes delivery of media content such as Video, Audio, Image, Text, Software, and Video Games over...
Performance Analysis of Cognitive Radio Networks With Capability of Spectrum Mobility
, M.Sc. Thesis Sharif University of Technology ; Ashtinani, Farid (Supervisor)
Abstract
In cognitive radio networks, secondary users try to use empty channels that primary users provide for transmitting data. First, the secondary users sense the target channels. If the selected channel is empty, they send their data. When the primary user returns, the secondary user has to find another empty channel for transmitting data. The process of changing channel is called hand off. In recent years, there are challenges in the spectrum mobility of for reducing delay time of secondary users packet. In this thesis, we assume that the secondary users network is an ad-hoc network. We use contestation based method for reducing collision between secondary users. In this thesis, we present an...
Energy Harvesting in Wireless Communication Networks
, Ph.D. Dissertation Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
In this dissertation, we focus on two types of EH cooperative networks. In the first type, the source and relay nodes access a shared channel according to a random access protocol. Also, they harvest energy from non-RF environmental resources. The relay node has data and energy buffers to store the packets of the source and the harvested energy, respectively. Our main goal is to find the optimal relaying protocol at the relay node. In this regard, the first type of the EH cooperative networks are investigated in two different scenarios. In the first one, the MAC channels of the nodes are independent of the applied policy at the relay node, and thus the throughput of the network is fixed....
Analysis of Age of Information in Multisource Communication Networks
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
In this thesis, we investigate the age of information in multisource networks. To this end, in the first phase of this thesis, we evaluate the Age of Information (AoI) in slotted ALOHA network. Regarding the interactions among the nodes in such a network, we model the memory of this network with a Trellis and find the peak Age of Information (PAoI) by using this Trellis. Then, we evaluate Age of Information-constrained regions and achievable throughput regions in the slotted ALOHA network and optimize Age of information in a symmetric case. In the second phase, we propose an efficient prioritizing scheme from AoI point of view and find the distribution of AoI and PAoI in such a scheme with...
Cooperative Networks from the MAC Layer Viewpoint
, Ph.D. Dissertation Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
Cooperative networks have attracted much attention in the last decade. Being introduced to mitigate the inefficiencies in the physical layer, cooperation can also be beneficial in upper layers, such as MAC layer. Using cooperative communications is specifically more challenging in the random access networks, where collision is a main concern. Delay, as one of the main performance metrics in random access networks, can decrease significantly by using cooperative communications. Transmission delay, i.e., the required time for delivering one packet form its source to destination, is an important metric in many communication networks, such as the networks in which each node generates and...
Latency Analysis and Improvement in Distributed Storage Systems
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
Today, with the proliferation of smart devices and the advent of various innovative applications, there has been an exponential surge in the number of users in telecommunication networks. Consequently, the data traffic rate is escalating considerably. This has created a major predicament for wireless networks due to the limited frequency spectrum. Therefore, the conventional centralized architecture makes it difficult for the network to address these numerous simultaneous requests. This issue can be solved by designing new architectures. In a way, network information is disseminated to the network's periphery and the contents of the primary data center are provided temporarily in closer...
Traffic Analysis in Cellular Networks based on OFDMA Access
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
OFDMA modulation method is a multiple access technique that is resistant to fading effect in cellular communication systems. In OFDMA, multiple access is performed both in time and frequency, so more users can be served with less delay than other protocols. In this thesis, based on queuing theory, we seek to present a traffic model for OFDMA system traffic analysis in single-cell mode and in the next section in multi-cell mode. Then, by examining and comparing the analytical results and the simulation results, we show that our proposed model for traffic analysis of the OFDMA system is a suitable and effective model. Also, since users in the multimedia environment need different types of...
Analysis and Reduction of the Age of Information in a Random Access Relay Network
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
In data transmission from a transmitter to a receiver, in some applications receiving updated data is crucial. To this end, a metric called Peak Age of Information has been introduced, the minimum of which is equivalent to having more updated packets. In this thesis, a method for calculating the probability distribution function of this random variable in a random access network with a transmitter, a relay, and a receiver in which the error probability of the channels between the transmitter and the relay and also the relay and the receiver is less than the error probability of the channel between the transmitter and the receiver, is proposed. The transmitter and the relay send their packets...
Analysis and Enhancement of Distributed Timely Computations Regarding Age of Information
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
Edge computing networks are ideal for emerging applications that require computation-intensive and timely processing, as they enable complex computation tasks to be offloaded and computed in parallel and distributed manner. However, for many applications that the calculations are of one type, such as determining the location of an object, it is crucial to insure the freshness of the information. Age of Information (AoI) metric is employed to measure freshness. This research focuses on identifying an appropriate model for calculating the average AoI in distributed computing systems and determining the optimal decision-making policy to enhance it. In essence, the primary objective is to...
Modeling Dependent Requests in Caching Mechanisms and Its Effect on Delay and Age of Information
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
In many communication scenarios, dependent requests are commonly observed, such as caching networks. In this thesis, we propose a new analytical model to examine the impact of dependent arrivals on the average response time in a simple queue with exponential service time. We examine two dependency scenarios: batch arrivals and the arrival of one customer after another with a random delay with a known distribution. To accurately model the effect of dependency between arrivals, we define redundant customers in our model to compensate for the existing dependency effects. We demonstrate the validity of our modeling approach analytically in the first scenario and through simulation in the second...
Reducing writebacks through in-cache displacement
, Article ACM Transactions on Design Automation of Electronic Systems ; Volume 24, Issue 2 , 2019 ; 10844309 (ISSN) ; Faraji, A ; Vakil Ghahani, S. A ; Samandi, F ; Lotfi Kamran, P ; Sarbazi Azad, H ; Sharif University of Technology
Association for Computing Machinery
2019
Abstract
Non-Volatile Memory (NVM) technology is a promising solution to fulfill the ever-growing need for higher capacity in the main memory of modern systems. Despite having many great features, however, NVM's poor write performance remains a severe obstacle, preventing it from being used as a DRAM alternative in the main memory. Most of the prior work targeted optimizing writes at the main memory side and neglected the decisive role of upper-level cache management policies on reducing the number of writes. In this article, we propose a novel cache management policy that attempts to maximize write-coalescing in the on-chip SRAM last-level cache (LLC) for the sake of reducing the number of costly...
Saturation Throughput Analysis of a Cognitive WLAN Overlaid on a Primary Network with Time-Scheduled Users
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
The activity status of primary users in a cognitive network scenario is a key parameter in saturation throughput of the secondary users. In this respect, usually a simple on-off traffic model with exponentially distributed intervals is exploited. For time-scheduled users in modern common standards like WiMAX or LTE this traffic model is not valid anymore. In this thesis, we focus on the saturation throughput analysis of a cognitive network scenario, comprised of a WLAN overlaid on a network with time-scheduled primary users like WiMAX. In this scenario, the cognitive nodes (CNs) hear downlink map (DL-MAP) and thus, know the frequency and time locations of all allocated slots at each downlink...
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
In this thesis, we propose an analytical model in order to evaluate the maximum stable throughput for unicast services in Vehicular Ad hoc NETworks (VANETs). In this respect, we employ two queueing networks in order to model the mobility patterns of the vehicles as well as the multi-hop packet transmission scenario. In the proposed model, we map the features of IEEE 802.11, as well as the details of multi-hop packet transmission, regarding the dynamic status of vehicles, onto suitable parameters of the queueing networks. In our modeling approach, we take the details of MAC and routing schemes into account by classifying the vehicles based on their mobility patterns and considering...
Throughput Analysis of Routing Algorithms in 802.11s-Based Wireless Mesh Networks
, M.Sc. Thesis Sharif University of Technology ; Ashtiani, Farid (Supervisor)
Abstract
Wireless Mesh Networks (WMNs) can be considered as the result of cooperation among two ad hoc networks, one composed of Mesh Points (MPs) and another one composed of Mesh Clients (MCs) as well as corresponding Access (Mesh) Points. Each one has its own functions and properties. This cooperation results in a flexible network that to fully demonstrate its potentials, requires special routing algorithms and access schemes.Here, we study and analyze three routing algorithms in IEEE 802.11s-based WMNs from throughput point of view. To this end, each node in the network is modeled by a queue during Contention-Free Periods (CFPs). During Contention-Based Period, however, to simplify modeling, we...
Fast data delivery for many-core processors
, Article IEEE Transactions on Computers ; Volume 67, Issue 10 , 2018 , Pages 1416-1429 ; 00189340 (ISSN) ; Lotfi Kamran, P ; Mazloumi, A ; Samandi, F ; Naderan Tahan, M ; Modarressi, M ; Sarbazi Azad, H ; Sharif University of Technology
2018
Abstract
Server workloads operate on large volumes of data. As a result, processors executing these workloads encounter frequent L1-D misses. In a many-core processor, an L1-D miss causes a request packet to be sent to an LLC slice and a response packet to be sent back to the L1-D, which results in high overhead. While prior work targeted response packets, this work focuses on accelerating the request packets. Unlike aggressive OoO cores, simpler cores used in many-core processors cannot hide the latency of L1-D request packets. We observe that LLC slices that serve L1-D misses are strongly temporally correlated. Taking advantage of this observation, we design a simple and accurate predictor. Upon...