Back Ground And Definesions for Delay Tolerant Network - Assignment Example

1. Describe what a DTN is (3 pages) a. What is a DTN? A DTN or Delay Tolerant Network is a wireless network (Abraham & Jebapriya, 2012, p.44) that does not have continuous network connectivity (Floreen et al, 2010, p.143). Its architecture includes a store-carry-and-forward paradigm using a bundle layer that provides internetworking on heterogeneous networks operating on different transmission media (Obaidat & Misra, 2011, p.102). The concept was first conceived for space communication impairments but later deployed to cover other networks particularly those that are vulnerable to long delays and link disruptions (Giambene & Sacchi, 2011, p.187). b. Describe Advantages and Challenges DTN has a number of advantageous properties not present in conventional networks. DTN offers opportunistic communication where sender and receiver communicate only a certain point in time. It is a sparse mobile network where nodes can be deployed over an unfamiliar environment but it can still work despite disruptions (Abraham & Jebapriya, 2012, p.44). This is because the absence of end-to-end connectivity, which traditionally prevents TCP or UDP connections, is not a problem with DTN “store-and-forward” techniques (Vasilakos et al, 2012, p.6). c. Provide example networks and applications of DTNs Networks in the DTN category include IPNl; Vehicular Networks such as DarkNet, Message Ferry, Village Network; MANNET an Ad Hoc Network such as Military tactical networks; Sensor Networks such as acoustic underwater networks; and Mule Networks including Zebranet for tracking zebras in wildlife, Sami Network Connectivity, Carrier Pigeons such as RFC 1149 & RFC 249 as implemented by Bergen Linux users group (Farahmand, 2008, p.1-42). d. Describe 3 routing DTN protocols (example epidemic routing, spray and wait) DTN routing protocol is divided in two classes- deterministic time evolving networks and stochastic (non-deterministic) time evolving networks. Common DTN routing protocols include epidemic, spray and wait. Epidemic routing occur when a node always try to send its message to neighbouring nodes. Sent messages are stored and tagged with a unique ID in the local buffer (Fathima & Wahidabanu, 2010, p.56). This DTN protocol is easier to implement and likely increase delivery success but sending messages this way can result to accumulation of redundant copies of a single message in a network. Similarly, such message sending behaviour can significantly waste network resources such as energy and buffer space (Balandin et al, 2010, p.277). Although a flooding-based routing similar to epidemic routing, Spray and Wait on the other hand limits the total number of copies of a single message to finite number “L”. In this routing protocol, nodes attempt to send limited copies of message to neighbouring nodes during the spray phase. In the Wait phase, each node retains one copy of the message and does not transfer until it finally found the destination of the message. Compared to epidemic, Spray and Wait, do not waste network resources as it limit packet transmission and wireless contention (ibid, 278; Thrasyvoulous et al, 2005, p.254). e. Attempt to find any data fusion work in DTNs Data fusion work in DTN include the tracking filter developed by Girija et al. (2000) in order to generate a unified data from multiple sensors. Another is the multi-sensor data fusion where three different messages from dissimilar type of sensors were fused to generate one valid sensor data. Fall et al. (2008) attempted “custody transfer” in DTN where fused messages are stored until it can be reliably delivered to another node (p.5). Similarly, Ramanathan et al. (2007), applied Delay Differentiated Gossiping or DDG in DTN where a base node is assigned as fusion centre in a sensor network. DTG detect node activities when a new packet is generated as well as when it receives a packet from another node (p.1-5). Wang (2007) deploys in his data centric routing in order to reduce implosion and overlapping problems. In his attempt, Yu Wang employed message fault tolerance and aggregation of data before they are transmitted to another node. These included designing a delivery scheme that utilizes erasure coding technology in order to achieve effective data delivery ration and low overhead (p.110). 2. Describe what RFIDs are (3 pages) a. Different types of RFIDs , passive and active, short range and long range RFID stands for Radio-frequency identification where the source of energy is the electromagnetic field generated by the antenna of the reader when it transmits. Electric current is induced whenever a tag is placed within this field and power up the chip. RFID technology can be found in commercial establishments and other industries being used for identification, tracking objects in a supply change, monitoring of object statues, and many more applications. An RFID system typically consists of one or more readers and a number of tags communicating to each other at a certain frequency and distance (Yenkataraman & Muntean, 2012, p.2). RFID reader or transceiver read the information from the RFID tags thus, it is usually connected to a back-end database. The reader has two key functional modules- HF or high frequency interface and the control unit. This interface can function as generator of transmission power to activate the tags, modulator of signals when sending request to RFID tags, and receiver and demodulator of signals sent by the tags. The control unit on the other hand perform a more complex function such execution of anti-collision algorithms particularly when communication with multi-tags, encryption and decryption, and authentication between RFID readers and RFID tags (Vacca, 2009, p.207). A passive RFID is a system where tags carry no battery and dependent on the reader for its energy requirements. Active RFID on the other hand comes with a batter in the tag, which they can use to transmit (Roussos, 2008, p.5). Active tags are more advantageous compared to passive RFIDs that derive all their power from the reader signal. First, an active tag can transmit better as it has more power and longer range. Since it has more power than passive tags, active RFIDs can work in hostile environment where radio frequencies are affected by electric machinery. However, despite power and range advantage, some users prefer passive tags as they do not depend on batteries that can expire and needs to be recharge or replaced in certain point in time (ibid, 5). Passive RFID tags have built in IC and antenna but no internal power supply. It is quite small and useful for short distance, which depending on the selected radio frequency and antenna design can be about 10 cm to a few metres. In contrast, also with IC and antenna, active RFID tags power the IC through own power source and more reliable than passive tag. Active RFID tags have longer range and larger memory but more expensive compared to passive RFID (Coskum et al, 2011, p.6). In terms of frequency, RFID use frequencies ranging from 300 kHz to 3 GHz and active tags transmit only at higher RF frequencies while passive tags can use all frequencies. For instance, passive RFID in low frequency band has a communication range of about 1 metre while 30 cm at high frequency. In ultra high frequency and microwave, both passive and active RFID tags are capable of transmitting up to 10 metres. However, some high-end active RFID can even go beyond 10 metres (ibid, 7). RFID systems that can communicate beyond 1 metre are considered long-range systems, operate using electromagnetic waves in the UHF, and microwave range. Majority of long-range systems are called backscatter systems because they have a different physical operating principle. Other long-range systems use surface acoustic wave transponders in microwave range. They are operated at 868 MHz (Europe) and 915 MHz (USA) in UHF and 2.5 GHz and 5.8 GHz at the microwave frequencies (Finkenseller, 2010, p.22). In contrast, short-range passive readers have at least three common frequencies; two-low frequency bands at 1254 and 134.2 KHz; and high-frequency readers at 13.56 MHz that allows fast read rates and longer-range reading distances (Igoe, 2007, p.306). RFID can also be classified in the terms of memory as some of them have read-only, write-once/read-may tags, and fully rewriteable tags. Similarly, in terms of computational power, they can be classified into three categories –basic, symmetric-key, and public-key tags. Basic tags cannot perform cryptography computation like public-key do while symmetric tags as the name suggest is limited to computation of symmetric key (Vacca, 2009, p.206). According to John Vacca, RFID tags can be further classified in terms of functionality such as Class 0 up to Class 4 defined by MIT Auto-ID Center. Class 0 tags are passive that do not contain memory and good only for electronic article surveillance. Class 1 tags are also passive but they have read-only or write-once/read-many memory thus useful in identification. Class 2 tags on the other hand are semi-active and active and they have rewritable memory that can offer data-logging functionality. Class 3 tags are also semi-active and active tags but they have on-board sensors that are useful in recording temperature, acceleration, motion, radiation and so on. Finally, Class 4 are active tags with fully rewritable memory and wireless networking components (ibid, 207). 3. Describe data fusion (3 pages) a. Define the term with respect to sensor networks (1 paragraph) A sensor network typically has battery operated nodes and a base station. These nodes are equipped with integrated sensors and capable of data process and short-range radio communication. Since battery-powered nodes have limited power and shorter communication range, they typically perform in-network data fusion (Khan, 2004, p.3). Data fusion occurs when these nodes gather results from different nodes and depending on the decision criteria being use combine these results with their own and send the fused data to another node or base station (ibid, p.4). b. Describe some early work on data fusion Richer data set can be acquired through fusion of multiple data sources and for this reason, much early research focus on the beneficial effects of data fusion. For instance, Mongi and Rafael in 1992 conducted a study about data fusion and found that sensory data fusion can improve signal quality (Hu, 2008, p.10). In early 2000, Hall & McMullen (2004), developed mathematical techniques in multi-sensor data fusion that include the JDL or Joint Directors of Laboratories data fusion process in five different levels. First, level include data fusion to determine position, velocity, attributes, and so on while the second level deals with situational assessment such relationships of observed entities. Level three on the other hand deals with projection of the current situation while level four deals with monitoring and subsequent refinement of the data fusion process. Finally, level 5 deals with the interaction between the data fusion system and human decision maker to further improve interpretation of results and decisions (p.3). Data fusion was already applied in area of defence such automated target recognition, battlefield surveillance, guidance and control of autonomous vehicles. According to Bhowmik et al, (2006), there are several techniques developed using data fusion such as J. Czyz et al, “Decision Fusion for Face Recognition” and B. Gokberk et al. “Decision Level Fusion for 3D Face Recognition” (p.2). These techniques include Discriminant Analysis or LDS and Support Vector Machine or SVM based algorithm. Here, data fusion produces an illumination-variant face image by detecting the eyeglass and replaced them with an eye template from visual and thermal face images (ibid p.2). c. Describe advantages and limitations There are a number of qualitative and quantitative benefits in multi-sensor data fusion and according to Hall & McMullen (2004), it can deliver robust operational performance as one sensor can contribute information if others are unavailable or jammed. It allows continuous operation despite jamming and increase probability of detection (p.22). It is capable of extended spatial and temporal coverage as one sensor can detect or measure a target when other sensor cannot. It can increase confidence as it applies rules of engagement that require positive target identification. It reduces ambiguity as fused information reduces hypotheses about the event (p.23). However, data fusion has its disadvantage in operational performance as it can increase the complexity of observations, as too much data may be confusing. In terms of extended spatial coverage, improper results may be gathered due to misaligned data while effectiveness of extended temporal coverage may be compromise if it cannot combine multimodal data that is not consistent in space and time with no physical feature correlation. Similarly, increased confidence may not be achieved if the result is just a number such as p (id) = 90% (Shahbazian et al, 2005, p.18). d. When there is mobility and no mobility Traditional routing protocols requires constant connectivity thus it is likely to suffer where there is node mobility as it limits the contact between nodes (Abraham & Jebapriya, 2012, p.44; Fathima & Wahidabanu, 2010, p.57). Similarly, data fusion in this type of routing would fail as data aggregation may not occur in address-centric routing (finding the shortest paths between pairs of end nodes) particularly when data is coming from wireless communication devices (Stojmenovic, 2005, p.495). Nodes in DTN are mobile and connectivity is only established when nodes are within range of each other (Abraham & Jebapriya 2012, p.44). Therefore, data fusion can work even if there is node mobility as it would have time locate routes that lead to the largest degree of data aggregation. e. What are the power constraints impact on fusion Data fusion in wireless sensor network requires data from sensors to be aggregated thus subject to delay. Similarly, data fusion performance may be negatively affected by the need to conserve power during data processing and communication in order to enhance network lifetime and survivability. For instance, saving power and bandwidth during data transmissions may be done through compression or quantization but resulting distortion will greatly affect the performance of data fusion (Wu et al, 2006, p.713). References: Abraham A. & Jebapriya S, (2012), Routing Strategies in Delay Tolerant Networks: A Survey, International Journal of Computer Applications, Vol. 42, No. 19, pp. 45-48 Balandin S, Dunaytsev R, & Koucheryavy Y, (2010), Smart Spaces and Next Generation Wired/Wireless Networking, Springer, Germany Bhowmik M, Majumder G, Bhattacharhjee D, (2006), Next Level of Data Fusion for Human Face Recognition, Jadavpur University, pp.1-8 Coskun V, Karem O, Ozdenizci B, (2011), Near Field Communication: From Theory to Practice, John Wiley & Sons, US Hall D. & Llinas J, (1996), An Introduction to Multisensor Data Fusion, Proceedings of the IEEE Vol. 85, No. 1, pp. 6-23 Khan A, (2004), Data Fusion in Sensor Networks, University of Buffalo, pp. 1-41 Fall K, Hong W, & Madden S, (2008), Custody Transfer for Reliable Delivery in Delay Tolerant Networks, Intel Research, pp.1-6 Farahmand F, (2008), Delay Tolerant Networks: Challenges and Applications, Central Connecticut University, p. 1-42 Fathima G. & Wahidabanu R, (2010), A New Queuing Policy for Delay Tolerant Networks, International Journal of Computer Applications, Vol. 1, No. 20, pp. 56-59 Finkenzeller K, (2010), RFID Handbook: Fundamentals and Applications in Contactless Smart Cards, Radio Frequency Identification and Near-Field, John Wiley & Sons, US Floreen P, Kruger A, & Spasojevic M, (2010), Pervasive Computing: 8th International Conference, Springer, Germany Giambene G. & Sacchi C, (2011), Personal Satellite Services: 3rd International ICST Conference, Springer, Germany Girija G, Raol J, Raj R, & Kashyap S, (2000), Tracking Filter and Multi-sensor Data Fusion, Sadhana, Vol. 25, pp.159-167 Igoe T, (2007), Making Thinks Talk: Practical Methods for Connecting Physical Objects, O’Reilly Media Inc., UK Hall D. & McMullen S, (2004), Mathematical Techniques in Multisensor Data Fusion, Artec House, US Hu J, (2008), Data Fusion: A First Step in Decision Informatics, ProQuest, US Misra S, Woungang I, & Misra S.C, (2009), Guide to Wireless Sensor Networks, Springer, Germany Obaidat M. & Misra S, (2011), Cooperative Networking, John Wiley & Sons, US Ramanathan P. & Singh A, (2007), Delay-Differentiated Gossiping in Delay Tolerant Networks, University of Wisconsi-Madison, pp.1-5 Qian Z., Cao L, & Su W, (2012), Recent Advances in Computer Science and Information Engineering, Springer, Germany Roussos G, (2008), Networked RFID: Systems, Software, and Services, Springer, Germany Shahbazian E, Rogova G, & Valin P, (2005), Data Fusion for Situation Monitoring, Incident Detection, Alert and Response Management, IOS Press, US Stojmenovic I, (2005), Handbook of Sensor Networks: Algorithms and Architectures, John Wiley & Sons, US Thrasyvoulos S, Psounis K, & Raghavendra C, (2005), Spray and Wait: An Efficient Routing Scheme for Intermittently Connected Mobile Networks, SIGCOMM Workshops, pp. 252-259 Vacca R, (2009), Computer and Information Security Handbook, Morgan Kaufmann Publishing, UK Vasilakos A., Zhang Y, & Spyropoulos T, (2012), Delay Tolerant Networks: Protocols and Applications, CRC Press, US Venkataraman H. & Muntean G, (2012), Green Mobile Devices and Networks: Energy Optimization and Scavenging Techniques, CRC Press, US Wang (2007), Delay/Fault-Tolerant Mobile Sensor Networks: A New Paradigm for Pervasive Information Gathering, University of Louisiana, pp. 1-136 Wu X. & Tian Z, (2006), Optimized Data Fusion in Bandwidth and Energy Constrained Sensor Networks, IEEE, ICASSP, pp.713-716 Read More