AX.25 packet radio is a popular means of data communications which has found many wide spread applications. For error control, AX.25 packet radio currently uses the go-back-N ARQ scheme which makes no attempt to correct errors, but only re-transmits packets which have been received with errors. This can lead to a very inefficient system when the channel error rate becomes substantially large. It has been found that by adding forward error correction (FEC) to the packet, the system performance can be substantially improved. This thesis studies the performance of various BCH codes, the (23,12) Golay code, Reed-Solomon codes, and different rate convolutional codes with varying constraint lengths when used in conjunction with the go-back-N ARQ. Code combining and concatenation are also studied. The performance of these codes is based on throughput performance, code rate, and system complexity. It is found that the (255,187) Reed-Solomon and the 1/2 rate v=9 convolutional codes can greatly enhance the performance of the AX.25 packet radio system. These codes provide good throughput performance over a large range of bit error rates and both are readily implemented. In conclusion, error control codes should be included with the AX.25 packet radio in order to improve its performance over noisy channels. / Master of Science

Modern networks carry a wide range of different data types, each with its own individual requirements. The scheduler plays an important role in enabling a network to meet all these requirements. In wired networks a large amount of research has been performed on various schedulers, most of which belong to the family of General Processor Sharing (GPS) schedulers. In this dissertation we briefly discuss the work that has been done on a range of wired schedulers, which all attempt to differentiate between heterogeneous traffic. In the world of wireless communications the scheduler plays a very important role, since it can take channel conditions into account to further improve the performance of the network. The main focus of this dissertation is to introduce schedulers, which attempt to meet the Quality of Service requirements of various data types in a wireless environment. Examples of schedulers that take channel conditions into account are the Modified Largest Weighted Delay First (M-LWDF), as well as a new scheduler introduced in this dissertation, known as the Wireless Fair Largest Weighted Delay First (WF-LWDF) algorithm. The two schemes are studied in detail and a comparison of their throughput, delay, power, and packet dropping performance is made through a range of simulations. The results are compared to the performance offour other schedulers. The fairness ofM-LWDF and WFLWDF is determined through simulations. The throughput results are used to establish Chernoff bounds of the fairness of these two algorithms. Finally, a summary is given of the published delay bounds of various schedulers, and the tightness of the resultant bounds is discussed. / Thesis (M.Sc. Eng.)-University of Natal, Durban, 2003.

An ad-hoc network is a collection of mobile nodes (wireless communication devices) that transmit data over systems that do not require any centralized control, such as that found in cellular networks. This makes ad-hoc networks suitable for military type applications, since there is no need for an established backbone infrastructure and hence no single-point-of-failure. However, other uses of ad-hoc systems include search and rescue missions, law enforcement operations, commercial and educational communication of laptop (and other handheld device) data, as well as in the transmission of environmental sensor information. The mobile ad-hoc concept brings many design challenges. The dynamic freedom of movement from mobile nodes causes random, sometimes rapidly time changing topologies, which are inappropriate for use through traditional wired protocols. In addition, wireless networks generally contain greater bandwidth, processing and power constraints than their wired counterparts, since they are implemented on embedded mobile, handheld devices. Thus, a different approach is needed in the wireless network domain. This has resulted in wireless routing protocols employing adaptive, multi-hop, distributed methodologies in which each node additionally acts as a router for each of its neighbouring nodes, in order to achieve a large degree of network connectivity. However, due to the broadcast nature of wireless transmissions, ad-hoc systems contain a point-to- multipoint communication architecture, making it well suited to multi-path traffic. One such application is in multicasting, which sends data from one source to two (or more) destinations. But, due to the shared characteristics of the communication channel, such traffic may cause multiple contentions and collisions to occur, which will degrade the efficiency and performance of a protocol. This dissertation examines these different design tradeoffs through the use of a freely available simulation package, known as NS-2 (Network Simulator - version 2). In addition, a novel routing protocol, known as LAMP (Location Aided Multicasting Protocol), is developed to handle time-bounded audio information, which is employed in a network that consists of sixteen commercial handheld devices. LAMP utilizes a destination-sequenced, next-hop routing table to forward multicast data. Since mobility causes neighbouring nodes to continually change, next-hop links need to be periodically updated. But, between each update period, a next-hop link may become broken. Thus, if a packet is required to be routed, for which its' next-hop link is unknown, LAMP reverts to a localized location aided flood to find a path to that destination. However, since flooding causes network congestion, it is only employed when its' table forwarding scheme fails. Results have shown that LAMP improves packet delivery ratios by up to 5% over exisiting flood-limiting schemes: Furthermore, LAMP has been shown to be comparable to leading schemes, even when employed to route data to a single source-destination pair. / Thesis (M.Sc.Eng.)-University of Natal, Durban, 2003.

Thesis ( M.Sc. (Computer Science)) -- University of Limpopo, 2019 / Cognitive radio network, which enables dynamic spectrum access, addresses the shortage of radio spectrum caused by ever-increasing wireless technology. This allows efficient utilisation of underutilised licenced spectrum by allowing cognitive radios to opportunistically make use of available licenced spectrum. Cognitive radios (CR), also known as secondary users, must constantly sense the spectrum band to avoid interfering with the transmission of the licenced users, known as primary users. Cognitive radios must cooperate in sensing the spectrum environment to avoid environmental issues that can affect the spectrum sensing. However, cooperative spectrum sensing is vulnerable to Byzantine attacks where selfish CR falsify the spectrum reports. Hence, there is a need to design and implement a defence mechanism that will thwart the Byzantine attacks and guarantee correct available spectrum access decisions. The use of extreme studentized deviate (ESD) test together with consensus algorithms are proposed in this study to combat the results of the availability of Byzantine attack in a cognitive radio network. The ESD test was used to detect and isolate falsified reports from selfish cognitive radios during the information sharing phase. The consensus algorithm was used to combine sensing reports at each time k to arrive at a consensus value which will be used to decide the spectrum availability. The proposed scheme, known extreme studentized cooperative consensus spectrum sensing (ESCCSS), was implemented in an ad hoc cognitive radio networks environment where the use of a data fusion centre (DFC) is not required. Cognitive radios make their own data fusion and make the final decision about the availability of the spectrum on their sensed reports and reports from their neighbouring nodes without any assistance from the fusion centre. MATLAB was used to implement and simulate the proposed scheme. We compared our scheme with Attack-Proof Cooperative Spectrum Sensing to check its effectiveness in combating the effect of byzantine attack.

Abstract available in digital copy. / Thesis (M.Sc.Eng.)-University of Natal, Durban, 2003.

Concurrent cooperative transmission (CCT) occurs when a collection of power-constrained single-antenna radios transmit simultaneously to form a distributed multi-input and multi-output (DMIMO) link. DMIMO can be a means for highly reliable and low-latency cooperative routing, when the MIMO channel is exploited for transmit and receive diversity; in this context, the range extension benefit is emphasized. Alternatively, DMIMO can be a means for high-throughput ad hoc networking, when the MIMO channel is used with spatial multiplexing. In both cases, concatenated DMIMO links are treated. The key contribution of this dissertation is a method of pre-synchronization of distributed single-antenna transmitters to form a virtual antenna array, in the absence of a global clock, such as a global positioning system (GPS) receiver or a network time protocol (NTP) to provide reference signals for the synchronization. Instead, the reference for synchronization comes from a packet, transmitted by the previous virtual array and simultaneously received by all the cooperative transmitters for the next hop. The method is realized for two types of modulation: narrowband non-coherent binary frequency-shift keying (NCBFSK) and wideband orthogonal frequency division multiplexing (OFDM). The pre-synchronization algorithms for transmission are designed to minimize the root-mean-square (RMS) transmit time, sampling and carrier frequency error between cooperative transmitters, with low implementation complexity. Since CCT is not supported by any existing standard or off-the-shelf radios, CT must be demonstrated by using software-defined radios (SDRs). Therefore, another contribution is a fully self-contained and real-time SDR testbed for CCT-based networking. The NCBFSK and OFDM systems have been designed and implemented in C++ and Python programming languages in the SDR testbed, providing practical performance of the CCT-based systems.

No description available.