Bluetooth Low Energy (BLE) has become a popular Internet of Things (IoT) technology. However, it was originally designed to only support the star topology. This PhD thesis investigates and evaluates different Bluetooth Low Energy (BLE) mesh network approaches, including existing ones (such as the Bluetooth Mesh standard), and our own solution for IPv6-based BLE mesh networking (6BLEMesh). The thesis comprises 6 main contributions:
1.- A comprehensive survey on existing BLE mesh networking proposals and a taxonomy for BLE mesh network solutions.
2.- An energy consumption model for Bluetooth Mesh. The model allows to predict useful performance parameters, such as device average current consumption, device lifetime and energy efficiency, considering the impact of the most relevant Bluetooth Mesh parameters, i.e. PollTimeout and ReceiveWindow, as well as application parameters (e.g. the data interval for a sensor that periodically reports its readings).
3.- A new proposed IPv6-based BLE mesh networking IETF standard (in progress), called 6BLEMesh. After defining the characteristics and properties of 6BLEMesh, we evaluated it in terms of connectivity, latency, RTT, and energy consumption.
4.- For the connectivity evaluation of 6BLEMesh, we developed an analytical model that takes a set of network and scenario characteristics as inputs, and provides two main results: i) the probability of no isolation of a node, and ii) the k-connectivity of the considered network. We validated the model by simulation.
5.- An implementation, and an experimental evaluation, of 6BLEMesh. We built a three-node testbed consisting of all node types (i.e. 6LN, 6LR and 6LBR). We used three different popular commercial hardware platforms. We evaluated a number of performance parameters on the testbed, related with latency and energy consumption. Next, we characterized the current consumption patterns of the complete life cycle for different node types in the three-node testbed. We also evaluated the energy performance of a 6LN on three different platforms. We presented a 6LN current consumption model for different connInterval settings. To this end, we experimentally characterized each current consumption state in terms of its duration time and average current consumption value. We illustrated the impact of connInterval on energy performance.
6.- A comparison between Bluetooth Mesh and 6BLEMesh, in terms of protocol stack, protocol encapsulation overhead, end-toend latency, energy consumption, message transmission count, end-to-end reliability, variable topology robustness and Internet connectivity. Bluetooth Mesh and 6BLEMesh offer fundamentally different BLE mesh networking solutions. Their performance depends significantly on their parameter configuration. Nevertheless, the following conclusions can be obtained. Bluetooth Mesh exhibits slightly greater protocol encapsulation overhead than 6BLEmesh. Both Bluetooth Mesh and 6BLEMesh offer flexibility to configure per-hop latency. For a given latency target, 6BLEMesh offers lower energy consumption. In terms of message transmission count, both solutions may offer relatively similar performance for small networks; however, BLEMesh scales better with network size and density. 6BLEMesh approaches ideal packet delivery probability in the presence of bit errors for most parameter settings (at the expense of latency increase), whereas Bluetooth Mesh requires path diversity to achieve similar performance. Bluetooth Mesh does not suffer the connectivity gaps experimented by 6BLEMesh due to topology changes.
Finally, 6BLEMesh naturally supports IP-based Internet connectivity, whereas Bluetooth Mesh requires a protocol translation gateway.
Bluetooth Low Energy (BLE) ha esdevingut una tecnologia popular per a Internet of Things (loT). Ara bé, va ser originalment dissenyada per suportar només la topologia en estrella. Aquesta tesi doctoral investiga i avalua diferents alternatives de xarxa mesh BLE, incloent alternatives existents (com l'estandard Bluetooth Mesh), i la nostra propia solució basada en IPv6, 6BLEMesh. Aquesta tesi comprén 6 contribucions·principals: 1.- Una revisió exhaustiva de l'estat de l'art i una taxonomia de les xarxes mesh BLE. 2.- Un model de consum d'energia per Bluetooth Mesh. El model permet predir parametres de rendiment útils, tals com consum de corrent, temps de vida del dispositiu i eficiéncia energética, considerant !'impacte deis principals parametres de Bluetooth Mesh (PollTimeout i ReceiveWindow) i a nivell d'aplicació. 3.- Un nou estandard (en progrés) anomenat 6BLEMesh. Després de definir les característiques de 6BLEMesh, aquesta solució ha estat avaluada en termes de connectivitat, laténcia, RTT i consum d'energia. 4.- Per a l'avaluació de connectivitat de 6BLEMesh, hem desenvolupat un model analític que proporciona dos resultats principals: i) probabilitat de no arllament d'un node i ii) k-connectivitat de la xarxa considerada. Hem validat el model mitjani;:ant simulació. .- Una imP.lementació, i una avaluació experimental, de 6BLEMesh. S'ha construrt un testbed de tres nodes, que comprén 5tots els tipus de node principals (6LN, 6LR i 6LBR). S'han usat tres plataformes hardware diferents. S'han avaluat diversos parametres de rendiment en el testbed, relacionats amb laténcia i consum d'energia. A continuació, s'ha caracteritzat els patrons de consum de corren! d'un ciclde de vida complet per als diferents tipus de nodes en el testbed. També s'han avaluat les prestacions d'energia d'un 6LN en tres plataformes diferents. S'ha presenta! un model de consum de corren! d'un 6LN per a diferents valors de connlnterval. Per aquest fi, s'ha caracteritzat emplricament cada estat de consum de corrent en termes de la seva durada i consum de corrent. 6.- Una comparativa entre Bluetooth Mesh i 6BLEMesh, en termes de pila de protocols, overhead d'encapsulament de protocol, laténcia extrem a extrem, consum d'energia, nombre de missatges transmesos, fiabilitat extrem a extrem, robustesa davant de topologies variables, i connexió a Internet. Bluetooth Mesh i 6BLEMesh són solucions de BLE mesh networking fonamentalment diferents. Les seves prestacions depenen de la seva configuració de parametres. Ara bé, es poden extreure les següents conclusions. Bluetooth Mesh mostra un overhead d'encapsulament de protocol lleugerament superior al de 6BLEmesh. Tots dos, Bluetooth Mesh i 6BLEMesh, ofereixen flexibilitat per configurar la laténcia per cada salt. Per un target de laténcia doni¡it, 6BLEMesh ofereix un consum d'energia inferior. En termes de nombre de missatges transmesos, les dues solucions ofereixen prestacions relativament similars per a xarxes petites. Ara bé, 6BLEMesh escala millor amb la mida i la densitat de la xarxa. 6BLEMesh s'aproxima a una probabilitat d'entrega de paquets ideal en preséncia d'errors de bit (amb un increment en la laténcia), mentre que Bluetooth Mesh requereix diversitat de caml per assolir unes prestacions similars. Bluetooth Mesh no pateix els gaps de connectivitat que experimenta 6BLLEMesh a causa de canvis en la topología. Finalment, 6BLEMesh suporta de forma natural la connectivitat amb Internet basada en IP, mentre que Bluetooth Mesh requereix un gateway de traducció de protocols.
