Manivasakan Rathinam

With the emergence of IoT, the number of devices that is getting connected is increasing exponentially, which poses a constraint on the lower latency requirements of processing these tasks. Evolving technologies such as fog computing and edge computing consist of computationally lesser exhaustive power servers, which bring this processing near the edge or onto the devices, thereby reducing the round-trip delay as well as the load on the entire network. In this work, we propose the hierarchical arrangement of servers in the fog computing layer for this processing of data and derive the analytical framework for the same. The proposed architecture works on top-down scheduling policy rule for the incoming packets, defined mathematically in terms of two-dimensional Markov chains. The performance of the proposed architecture is compared with an equivalent flat architecture analytically in terms of the mean sojourn time and the mean computational power and justified using simulation results.

We propose and demonstrate a generalized framework for performance limit evaluation and comparison of multihop optical repeated and regenerated links. The model developed is implementation agnostic and applies to multihop optical links of varied forms, including fiber, free space, and underwater links. The framework estimates the best-case performance gains of deploying an all-regenerative link over an all-repeater link for any given implementation. The implementation-independent technique is then illustrated using guided and free-space optical links. An abstract model is developed first with the evolution of signal, noise power, and bit error rate down the link compared and contrasted for both cases. The model is then evaluated using physical parameters for a typical fiber optic intensity-modulated direct detection link, and the obtained all-regenerator link performance advantage is translated to extra reach and lower transmission power requirements. Further, certain approximations are provided to reduce computational complexity and improve the analytical tractability of the procedure, which could be particularly helpful when employed in specialized hardware or for dynamic reconfiguration networks. Finally, the framework's versatility is established by employing it in analyzing an ideal free-space link and comparing amplify and forward links against decode and forward counterparts. Similar results are also reproduced on a commercial optical link simulation suite. Detailed literature on link analysis is provided for fiber, free space, and underwater links, bringing out their similarities. We conclude by elaborating on various current and emerging application domains and certain limitations of the proposed technique.

The advent of optical amplifiers have revolutionized optical communications, facilitating (Wavelength Division Multiplexed) WDM systems and leading to an explosion in bandwidth and data rates. But long amplified links degrades the optical Signal to Noise Ratio (SNR) with each stage of amplification due to the addition of Amplified Spontaneous Emission (ASE). This is overcome by using regenerators which does the Optical-Electrical and back to Optical (OEO) conversion to detect the signal and regenerating it anew thereby removing all the noise and hence restoring the SNR. But this electrical conversion introduces delays and prohibitively increases the cost of the link. A potential solution for this is the use of all optical regenerators, which does the amplification and reshaping in the optical domain itself. This paper undertakes a BER analysis proving the unconditional superiority of regenerators over repeaters for given link parameters. It also demonstrates the better power savings and extra reach of all optical regenerators. Further, a few approximations are used to represent the Q function analytically and the corresponding expressions are used for comparing the performance of repeaters and regenerators.

In many communication networks, controlling the packet delay variation (pdv) (in packet switched networks (PSN)) or jitter (in TDM networks) are challenging. In PSNs, pdv is crucial for real-time (e.g. telephony) and near real-time (e.g. streaming audio, video, VoD). Conventionally, many methods were adopted among which the most primitive way of traffic regulation is through the leaky bucket algorithm wherein the bursty sources in the internet are 'regulated' or 'shaped'. There are other schemes which are adopt 'deterministic' way to reduce jitter. In this paper, we adopt a method of dictating the service distribution (marginal joint) to control the output pdv or jitter. We rely heavily on the result in [1] which showed that positive correlations in service intervals do reduce the mean variance of queue length while it is vice versa for negative correlations. We show analytically that for negatively correlated queue, the IDI variance decreases as the correlation increases (and vice versa for positive correlation). But this is at the cost of increased delay (waiting time plus service time). Note that IDI variance appears as pdv or jitter at receiver/downstream node. To demonstrate the efficacy of the above findings, we apply the above to steaming audio and TDM over PSN problems. It is demonstrated that the jitter is minimized at the receiver.

Long haul optical links provides the backbone for most of today's communication and the Internet in general. The throughput of an optical link witnessed a major breakthrough with the advent of optical amplifiers in the early 1990s. These optical amplifiers with their large bandwidth facilitated the introduction of Wavelength Division Multiplexing (WDM) and a bandwidth explosion followed. Though they provide good gain and large bandwidth, all while keeping the information in optical domain; optical amplifiers do add noise of its own to the data, which when cascaded over long distances starts to limit the link length. Optical-Electrical-Optical (OEO) regenerators are used to clean and regenerate the signal but are expensive and adds potential delays in the link. In this paper we investigates their theoretical performance differences and highlights the advantages of all regenerators link over all repeaters ones. We undertake this analytical study to investigate the absolute theoretical gains of implementing an all regenerator link which is to serve as a baseline, or a precursor, for further investigations on the advantages of all optical regenerative link. Here we derive the performance limits of an all regenerator systems and compare it with its amplifier/repeater counterparts. Noise evolution in all repeater links and BER accumulation for both are illustrated. We illustrate general optical link BER curves and compare all repeater/regenerator link performances against input power for single and multiple hops. We also translate the BER advantage of all regenerator link to longer link reaches or lower power requirements. The later is repeated for different target BER too.

Next-generation wireless networks are envisaged to provide end-users ubiquitous low-latency computing services using devices at the network edge and machines before reaching the core cloud network. The crucial concepts of driving this technology are to offload computationally intensive tasks from users to edge or fog devices. The mobile edge network consists of an access point (AP), Radio Access Network (RAN), devices, edge servers, fog servers and finally the core cloud network. Given that fog computing is an emerging technology, it is imperative to study the performance of such systems analytically. Though there are many papers on performance analysis, many of them provide heuristic, ad-hoc solutions or pose it as optimization problems. In this work, we model a hierarchical fog architecture using Markov chain analysis and compare it with a flat fog architecture by investigating the different performance measures analytically and by simulation.

Wireless communication failures are inevitable as far as the current wireless environment is concerned. A disruption between the mobile device and wireless network links could halt the live running applications in mobile. Motivated by these problems, we have used queueing theory to analyze the Delayed Offloading model's performance with Failures using Balking (DOFB). The balking scheme is characterized by current OffLoading System (OLS) availability and the number of jobs queued for transmission. The proposed scheme achieves a reduction in job loss and mean queueing delay. The performance of the introduced scheme is mathematically analyzed using matrix-geometric methods and verified through simulation.