abstract: Emerging from years of research and development, the Internet-of-Things (IoT) has finally paved its way into our daily lives. From smart home to Industry 4.0, IoT has been fundamentally transforming numerous domains with its unique superpower of interconnecting world-wide devices. However, the capability of IoT is largely constrained by the limited resources it can employ in various application scenarios, including computing power, network resource, dedicated hardware, etc. The situation is further exacerbated by the stringent quality-of-service (QoS) requirements of many IoT applications, such as delay, bandwidth, security, reliability, and more. This mismatch in resources and demands has greatly hindered the deployment and utilization of IoT services in many resource-intense and QoS-sensitive scenarios like autonomous driving and virtual reality. I believe that the resource issue in IoT will persist in the near future due to technological, economic and environmental factors. In this dissertation, I seek to address this issue by means of smart resource allocation. I propose mathematical models to formally describe various resource constraints and application scenarios in IoT. Based on these, I design smart resource allocation algorithms and protocols to maximize the system performance in face of resource restrictions. Different aspects are tackled, including networking, security, and economics of the entire IoT ecosystem. For different problems, different algorithmic solutions are devised, including optimal algorithms, provable approximation algorithms, and distributed protocols. The solutions are validated with rigorous theoretical analysis and/or extensive simulation experiments. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2019

Payment Channel Networks (PCNs) provide a solution to the scalability problem in blockchain technology. They facilitate multiple-hop transactions via payment channels between peers, allowing for the execution of several transactions before updating each node’s balance on the blockchain. However, the performance of the network in effectively routing payments is affected by unbalanced channels. This imbalance reduces the network’s ability to route payments in both directions within a channel, resulting in decreased overall performance. Previous research has identified unidirectional payment flows in the network as the underlying cause of this issue. The payment flow is based on the demand for payments between peers and the paths these payments take. The aim of this study is to investigate the relationship between payment patterns and the average channel imbalance, where the payment pattern is a combination of demand patterns and what pathfinding algorithm is used. The study uses a model of PCNs that allows the relationship of the variables of interest to be measured, while other impacting variables are controlled. A deterministic value of the average channel imbalance for a PCN topology, demand pattern, and pathfinding strategy is achieved by computing the steady-state imbalance. The results show that the less homogeneous the demand pattern is, the more asymmetric the payment flow becomes, and the more imbalanced the network’s channels become. The results confirm what previous research mentions: unidirectional flow causes the payment channel to become imbalanced. The demand pattern that generates unidirectional flow imbalance significantly affects the entire network, regardless of the pathfinding strategy employed. Even with the possibility of a pathfinding strategy influencing payment flow to make it less unidirectional, the inherent unbalanced nature of the demand pattern remains a considerable challenge. The results also show that the pathfinding strategy has a less significant impact than the demand pattern on the balance of the network over time. / Payment Channel Networks (PCN) tillåter fler transaktioner utföras på kortare tid för en mindre kostnad, än att publicera varje enskild transaktion direkt på blockkedja och är en bra lösning på skalbarhetsroblem kopplat till blockedjor som baseras på proof-of-work. PCNs möjliggör för flerhoppstransaktioner via noder som kan vidarebefodra flera betalningar på varje enskild betalningskanal de håller öppna med andra noder. Inte förrän en kanal stängs publiceras den senaste uppdateringen av saldot för enskild betalningskanal på blockkedjan. Trots detta lider nätverkets prestanda av att kanaler blir obalanserade och ökar risken för att transaktioner misslyckas routas. Tidigare forskning nämner att orsaken till att kanaler blir obalanserade är enriktade betalningsflöden i nätverket, vilket beror av betalningsflödet som är en kombination av betalningsefterfråga (vem vill betala vem) mellan noderna och vilka vägar som dessa betalningar tar. Syftet med denna studie är att undersöka sambandet mellan betalmönster och genomsnittlig kanalobalans, där betalmönstret är en kombination av efterfrågansmönster och vägvalsstrategier. Studien använder en PCN-modell som möjliggör att undersöka sambandet mellan de variabler som är tänkt ska mätas, samtidigt som andra variabler kan kontrolleras. Genom att beräkna den stabila obalansen uppnås ett deterministiskt värde för genomsnittlig kanalobalans i varje nätverkstopologi. Resultaten visar att ju mindre homogent efterfrågemönstret är, desto mer asymmetriskt blir betalningsflödet och desto mer obalanserade blir nätverkets kanaler. Resultatet bekräftar det som tidigare forskning nämner: enriktade flöden orsakar obalans i betalningskanalen. Specifikt är det efterfrågansmönstern som skapar enriktade flöden som har de högsta stabila obalansen, oavsett vilken vägvalsstrategi som används. Resultaten visar också att vägvalsstrategin har mindre betydelse än efterfrågemönstret, åtminstone för vägvalsstrategier som leder till liknande valda vägar.