Ενεργά λέμε τα δίκτυα τα οποία επεξεργάζονται και τα περιεχόμενα (και όχι μόνο την επικεφαλίδα) των πακέτων που μετάγουν. Επικεντρωνόμαστε στα ενεργά δίκτυα ενθυλάκωσης, όπου ο προς εκτέλεση κώδικας συμπεριλαμβάνεται στα μεταγώμενα πακέτα, σε αντιδιαστολή με τους προγραμματιζόμενους μεταγωγείς. Παρουσιάζεται αναλυτικά το Active Networks Encapsulation Protocol (ANEP). Παρουσιάζονται δικτυακές εφαρμογές στις οποίες τα ενεργά δίκτυα βελτιώνουν την απόδοση της εφαρμογής και ταυτόχρονα μειώνουν τις απαιτήσεις σε δικτυακούς πόρους. Ακολούθως εστιάζουμε στην \"Ενεργή Αξιόπιστη Πολλαπλή Μετάδοση\", ένα πρωτόκολλο αξιόπιστης πολλαπλής μετάδοσης το οποίο χρησιμοποιεί την ενεργή μεταγωγή για να διαχειριστεί την ανάκτηση απωλειών πακέτων εντός του δικτύου (καταστολή NACK, λανθάνουσα μνήμη για πακέτα διόρθωσης, πολλαπλή μετάδοση περιορισμένης εμβάλειας). Παρέχονται αποτελέσματα προσομοιώσεων που υποστηρίζουν την θέση ότι ακόμα και με μικρό ποσοστό ενεργών κόμβων, ένα ενεργό δίκτυο μπορεί να βελτιώσει ουσιαστικά τις επιδόσεις της εφαρμογής και να μειώσει ταυτόχρονα την χρήση εύρους ζώνης. Κλείνοντας, κάνουμε κάποια τελικά σχόλια και εξάγουμε συμπεράσματα σχετικά με το υψηλό κόστος εγκατάστασης και συντήρησης των ενεργών δίκτυων, και πως αυτό αντιδιαστέλλεται με τα πλεονεκτήματα των τελευταίων σε σχέση με τις επιδόσεις των εφαρμογών και την χρήση των δικτυακών πόρων. / Active Networks are networks consisting (at least partially) of active nodes. A node is active if it doesn’t only processes a packet’s header in order to route it, but is also able to evaluate and process the packet’s payload. There are two kinds of active networks, depending on whether they are based on programmable switches or on capsules which bundle code together with the data. This dissertation focuses on the latter. The operational model of an active network of this kind comprises code execution models, network node management models and resource allocation policies. The Active Networks Encapsulation Protocol (ANEP) sets the mechanism for defining the platform required to evaluate the code that is encapsulated in the packet, as well the nodes’ behaviour when they do not support the required platform (drop the packet or simply forward it). This mechanism provides active networks with the flexibility to operate even when a very small percentage of the network’s node is actually active. There are various situations and where active networks can make better use of network resources. There are, for example, applications where different users might make similar, but different, requests resulting in unnecessary bandwith consumption when supported by conventional caching mechanisms. Active networks can provide smart caches that will dynamically synthesize pages from data cached by previous requests. Another situation where active networks can improve network performance is network applications like tele-conference, that depend heavily on new network services. Active networks allow for the faster deployment of new network services that enhance network speed and security and rationalise bandwidth usage through, for example multicast. Furthermore, active networks can support specialised applications, like for example on-line auctions, with custom-made network services. It is important to note that when measuring network performance, one should focus onto the network application’s performance, rather than network per-packet metrics like throughput and latency. In other words, intranetwork processing might increase both packet size and latency, but will improve the application’s end-to-end performance and reduce total network load. The protocols for three innovative network applications are presented: active reliable multicast, auctions over the network and remote sensor merging. For each of these we present network services that can be easily implemented and deployed in active networks to improve application performance. Finally, a more detailed analysis (by means of simulation) of an active reliable multicast protocol is presented. Active networks achieve two ends: on the one hand they push the idea of a network proxy to its logical end by effectively turning all network elements into smart proxies that provide caching, filtering, NACK suppression and other services. On the other hand they carry out part of the computation inside the network, bringing it closer to the data sources. When the computation is, for example, data merging this is beneficial to both the application and the network resources. This, however, can only be achieved at a cost. First of all in hardware, since network elements have to be upgraded from simple routers to full-blown computers capable of supporting Java and scripting languages. But also in latency, since packets have to undergo much more complex processing along the way that simple routing. In the applications presented here the costs associated with active transport are counter-balanced by the advantages the latter has to offer to the application as well as to the network. The bet that active networks have to win in order to get widely accepted, is to have enough active application protocols developed that their installation and maintenance cost can be justified.

A mobile wireless ad hoc network (MANET) consists of a group of mobile nodes communicating wirelessly with no fixed infrastructure. Each node acts as source or receiver, and all play a role in path discovery and packet routing. MANETs are growing in popularity due to multiple usage models, ease of deployment and recent advances in hardware with which to implement them. MANETs are a natural environment for multicasting, or group communication, where one source transmits data packets through the network to multiple receivers. Proposed applications for MANET group communication ranges from personal network apps, impromptu small scale business meetings and gatherings, to conference, academic or sports complex presentations for large crowds reflect the wide range of conditions such a protocol must handle. Other applications such as covert military operations, search and rescue, disaster recovery and emergency response operations reflect the "mission critical" nature of many ad hoc applications. Reliable data delivery is important for all categories, but vital for this last one. It is a feature that a MANET group communication protocol must provide. Routing protocols for MANETs are challenged with establishing and maintaining data routes through the network in the face of mobility, bandwidth constraints and power limitations. Multicast communication presents additional challenges to protocols. In this dissertation we study reliability in multicast MANET routing protocols. Several on-demand multicast protocols are discussed and their performance compared. Then a new reliability protocol, R-ODMRP is presented that runs on top of ODMRP, a well documented "best effort" protocol with high reliability. This protocol is evaluated against ODMRP in a standard network simulator, ns-2. Next, reliable multicast MANET protocols are discussed and compared. We then present a second new protocol, Reyes, also a reliable on-demand multicast communication protocol. Reyes is implemented in the ns-2 simulator and compared against the current standards for reliability, flooding and ODMRP. R-ODMRP is used as a comparison point as well. Performance results are comprehensively described for latency, bandwidth and reliable data delivery. The simulations show Reyes to greatly outperform the other protocols in terms of reliability, while also outperforming R-ODMRP in terms of latency and bandwidth overhead.

IP multicasting is a method for transmitting the same information to multiple receivers over IP networks. Reliability issue of multicasting contains the challenges for detection and recovery of packet losses and ordered delivery of the entire data. In this work, existing reliable multicast protocols are classified into three main groups, namely tree based, NACK-only and router assisted, and a representative protocol for each group is selected to demonstrate the advantages and disadvantages of the corresponding approaches. The selected protocols are SRM, PGM and RMTP. Performance characteristics of these protocols are empirically evaluated by using simulation results. Network Simulator-2 (ns2), a discrete event simulator is used for the implementation and simulation of the selected protocols. The contributions of the thesis are twofold, i.e. the extension of the ns library with an open source implementation of RMTP which did not exist earlier and the evaluation of the selected protocols by investigating performance metrics like distribution delay and recovery latency with respect to varying multicast group size, network diameter, link loss rate, etc.

I avhandlingen beskrivs ett hypotetiskt system som kan användas av mobila användare, bland andra taxichaufförer, som exempelvis vill följa en viktig fotbollsmatch. Flera faktorer ställer till problem: Ibland står bilen still och föraren har inget annat att tänka på än matchen. Ibland kör denne runt med en kund som inte vill bli störd av matchen. Dessutom kan det vara svårt att titta på rörliga bilder och köra bil samtidigt. I och med att bilen körs runt har man också olika bra anslutning till Internet vid olika tillfällen – det kan variera mellan inget alls, en dålig GSM/GPRS förbindelse (8 kbps) och en snabb WLAN anslutning (100 Mbps). I avhandlingen presenteras en tre-lagers modell som kan användas för att beskriva den här typen av applikationers kvalitet. Modellen delas in i tre lager: nätverk, applikation och användare/använding. Det sistnämnda lagret ligger utanför det tekniska systemet och definieras av att det är där de verkliga informationsutbytet sker. På applikationsnivån samlas data in, packas och packas upp i samband med nätverkstransport och visas sedan för användaren. Det är också här som eventuell interaktion sker med användaren. Nätverkslagret är ansvarigt för ändamålsenlig transport av data. De tre lagren är ömsesidigt beroende av varandra. Dålig prestanda på ettlager påverkar de andra lagren och tvärtom. Tre studier har genomförts av hur problem på nätverkslagret i form av begränsad bandbredd och hög fördöjning påverkar användarna. Låg bandbredd ger låg videokvalitet vilket inte uppskattas av användarnamnen genom att skifta till animeringar som fungerar med lägre bandbredd kan man ändå få användarna nöjda. Om användarna måste välja mellandålig videokvalitet och animeringar väljer de som ser sig som fotbollskunniga det förstnämnda och de som ser sig som okunniga men dock fotbollsfans väljer det sistnämnda. Men i en annan studie där användarna spelade bluffstopp mot varandra över ett datanätverk fick vi ett annat resultat. Där var det negativt med högre videokvalitet (bilder per sekund). En förklaring kan vara att användarna distraherades mer av högre bildfrekvens. I den tredje studien studerades vad som händer i Pong om man läggerin fördröjningar i spelet. Sedan tidigare visste man att det blir svårare attspela med fördröjningar – särskilt om man inte märker dem. Vi ställde ossfrågan om man kan kompensera för dem genom att informera användarna om dem. Det visade sig att användare som får information med i vårtfall en prediktiv visning lättare anpassar sin mentala insats till uppgiftens svårighetsgrad. Det är alltså inte bara möjligt utan ibland också önskvärt att utnyttja en lägre bandbredd från användarens perspektiv. Med det sagt finns det ändå i långt fler situationer där det är bättre med bättre nätverksprestanda. Pongspelet var roligare med lägre delay. Videon uppfattades som bättre medhögre bandbredd i den förstnämnda studien. Multicast, där ett paket skickas till flera användare i stället för att de skafå varsin, identiska paket, är ett viktigt verktyg för att få bättre prestanda i videoapplikationer. Tyvärr är det inbyggda stödet för multicast i den viktiga IEEE 802.11 standardfamiljen för trådlösa nätverk mycket outvecklat. Ettstort problem är att det inte går att veta om ett paket har kommit fram eller om det har försvunnit i en, mycket trolig, krock. Vi har vidareutvecklat och anpassat en föga känd krockdetektionsmekanism från 80-talet för använding i IEEE 802.11 nätverk. Den anpassade algoritmen kallar vi EMCD vilket är en förkortning för ‘‘Early Multicast Collision Detection’’ eller tidig krockupptäckt för multicast. Vi har presenterat en nysannolikhetsbaserad modell för att beräkna algoritmens prestanda undermaximal belastning. Modellen som har verifierats genom simuleringar kanäven användas för att beräkna optimala parametrar för algoritmen. Algoritmen har visats kraftigt reducera risken för oupptäckta kollisioner och reducerar den tid som går åt för dem. EMCD-algoritmen inspirerade till att utveckla en ytterligare algoritm som inte bara kan upptäcka utan också undvika kollisioner: PREMA som står för ‘‘Prioritized Repeated Eliminations Multiple Access’’ eller prioriterad kanal-åtkomst med upprepade eliminationer. Det finns två viktiga skillnader mellanhur de fungerar. I EMCD bygger kollisionsdetektionen på rektangelfördelade slumptal och en enda upptäcktsomgång. I PREMA används i stället geometriskt fördelade slumptal och upprepade omgångar. Effekten blir att man med stor säkerhet får en enda vinnare. även för PREMA presenteras en sannolikhetskalkylsbaserad prestandaanalys för maxlastfallet vilken stöds av simuleringar. Samma formler kan användas för att approximativt skatta prestanda i EY-NPMA som är en närliggande algoritm. Den var tänkt att använda i Hiperlan/1; en standard som aldrig fick något kommersiellt genombrott. Använder man den modell som vi presenterar i avhandlingens sista studiekan man med ganska god noggrannhet beräkna optimala parametrar för EY-NPMA med en beräkningsinsats O(mY S) mot O(mES×mY S) för tidigare kända algoritmer. / Imagine a taxi driver wanting to watch a football game while working. Events in the game cannot be predetermined, the driver's available attentional resources vary and network connections change from non-existing to excellent, so it will be necessary to develop a viewing application that can adapt to circumstances. This thesis presents a system model and sketches a framework for design and run time adaptations. The model has three layers: user/usage, application and network. Quality of service metrics are proposed for each layer. A particular emphasis is placed on the difference between the user/usage layer and the application layer. Satisfaction at the former means a job well done, a match played to your liking etc. Satisfaction at the latter means good picture quality, nice colours etc. The thesis continues by identifying and describing elements required to build the system used by the taxi driver. Three studies are presented where either bandwidth or delay are varied at the network level. Video is better the higher the bandwidth; animations can be used as a complement. They are shown to be better than low quality video but worse than high quality video for watching a football game. Better video in the form of higher frame rates turned out to be worse for playing a card game over the Internet. A possible explanation is the distraction experienced when the image is updated constantly. Another result of our studies is that users can adapt their mental effort to the actual load when given feedback on the network delay affecting a computer game. The results mentioned above show that it is possible to compensate for poor network performance. For the user, improved network performance is generally more satisfactory. Early multicast collision detection is a method for improved multicast performance in high load IEEE 802.11 networks. Prioritised repeated eliminations multiple access is a method for multicast and other traffic which can be used alone or in an IEEE 802.11 network. Probabilistic performance analysis and simulations show that both protocols drastically reduce the time spent in collisions and improve throughput compared to IEEE 802.11. Some of the formulae are applied to EY-NPMA as well; they are used to estimate performance and to estimate optimal operating parameters more efficiently than with previously known methods.

Le transport multicast est une solution efficace pour envoyer le même contenu à plusieurs récepteurs en même temps. Ce mode est principalement utilisé pour fournir des flux multimédia en temps réel. Cependant, le multicast classique de l’IEEE 802.11 n'utilise aucun mécanisme d’acquittement. Ainsi, l’échec de réception implique la perte définitive du paquet. Cela limite la fiabilité du transport multicast et impact la qualité des applications vidéo. Pour résoudre ce problème, 802.11v et 802.11aa sont définis récemment. Le premier amendement propose Direct Multicast Service (DMS). D'autre part, le 802.11aa introduit GroupCast with Retries (GCR). GCR définit deux nouvelles politiques de retransmission : Block Ack (BACK) et Unsolicited Retry (UR).Dans cette thèse, nous évaluons et comparons les performances de 802.11v/aa. Nos résultats montrent que tous les nouveaux protocoles multicast génèrent un overhead de transmission important. En outre, DMS a une scalabilité très limitée, et GCR-BACK n'est pas approprié pour des grands groupes multicast. D’autre part, nous montrons que DMS et GCR-BACK génèrent des latences de transmission importantes lorsque le nombre de récepteurs augmente. Par ailleurs, nous étudions les facteurs de pertes dans les réseaux sans fil. Nous montrons que l'indisponibilité du récepteur peut être la cause principale des pertes importantes et de leur nature en rafales. En particulier, nos résultats montrent que la surcharge du processeur peut provoquer un taux de perte de 100%, et que le pourcentage de livraison peut être limité à 35% lorsque la carte 802.11 est en mode d’économie d'énergie.Pour éviter les collisions et améliorer la fiabilité du transport multicast, nous définissons le mécanisme Busy Symbol (BS). Nos résultats montrent que BS évite les collisions et assure un taux de succès de transmission très important. Afin d'améliorer davantage la fiabilité du trafic multicast, nous définissons un nouveau protocole multicast, appelé Block Negative Acknowledgement (BNAK). Ce protocole opère comme suit. L’AP envoi un bloc de paquets suivi par un Block NAK Request (BNR). Le BNR permet aux membres de détecter les données manquantes et d’envoyer une demande de retransmission, c.à.d. un Block NAK Response (BNAK). Un BNAK est transmis en utilisant la procédure classique d’accès au canal afin d'éviter toute collision avec d'autres paquets. En plus, cette demande est acquittée. Sous l'hypothèse que 1) le récepteur est situé dans la zone de couverture du débit de transmission utilisé, 2) les collisions sont évitées et 3) le terminal a la bonne configuration, très peu de demandes de retransmission sont envoyées, et la bande passante est préservée. Nos résultats montrent que BNAK a une très grande scalabilité et génère des délais très limités. En outre, nous définissons un algorithme d'adaptation de débit pour BNAK. Nous montrons que le bon débit de transmission est sélectionné moyennant un overhead très réduit de moins de 1%. En plus, la conception de notre protocole supporte la diffusion scalable de lavvidéo. Cette caractéristique vise à résoudre la problématique de la fluctuation de la bande passante, et à prendre en considération l'hétérogénéité des récepteurs dans un réseau sans fil. / The multicast transport is an efficient solution to deliver the same content to many receivers at the same time. This mode is mainly used to deliver real-time video streams. However, the conventional multicast transmissions of IEEE 802.11 do not use any feedback policy. Therefore missing packets are definitely lost. This limits the reliability of the multicast transport and impacts the quality of the video applications. To resolve this issue, the IEEE 802.11v/aa amendments have been defined recently. The former proposes the Direct Multicast Service (DMS). On the other hand, 802.11aa introduces Groupcast with Retries (GCR) service. GCR defines two retry policies: Block Ack (BACK) and Unsolicited Retry (UR).In this thesis we evaluate and compare the performance of 802.11v/aa. Our simulation results show that all the defined policies incur an important overhead. Besides, DMS has a very limited scalability, and GCR-BACK is not appropriate for large multicast groups. We show that both DMS and GCR-BACK incur important transmission latencies when the number of the multicast receivers increases. Furthermore, we investigate the loss factors in wireless networks. We show that the device unavailability may be the principal cause of the important packet losses and their bursty nature. Particularly, our results show that the CPU overload may incur a loss rate of 100%, and that the delivery ratio may be limited to 35% when the device is in the power save mode.To avoid the collisions and to enhance the reliability of the multicast transmissions, we define the Busy Symbol (BS) mechanism. Our results show that BS prevents all the collisions and ensures a very high delivery ratio for the multicast packets. To further enhance the reliability of this traffic, we define the Block Negative Acknowledgement (BNAK) retry policy. Using our protocol, the AP transmits a block of multicast packets followed by a Block NAK Request (BNR). Upon reception of a BNR, a multicast member generates a Block NAK Response (BNAK) only if it missed some packets. A BNAK is transmitted after channel contention in order to avoid any eventual collision with other feedbacks, and is acknowledged. Under the assumption that 1) the receiver is located within the coverage area of the used data rate, 2) the collisions are avoided and 3) the terminal has the required configuration, few feedbacks are generated and the bandwidth is saved. Our results show that BNAK has a very high scalability and incurs very low delays. Furthermore, we define a rate adaptation scheme for BNAK. We show that the appropriate rate is selected on the expense of a very limited overhead of less than 1%. Besides, the conception of our protocol is defined to support the scalable video streaming. This capability intends to resolve the bandwidth fluctuation issue and to consider the device heterogeneity of the group members.