Research On Cluster Channel Allocation In Wireless Industrial Network

With the fast development of IT, industrial wireless network technology has become a hotspot in the research of automation field, and has been a new, device-oriented information exchanging of industrial wireless telecommunication technology. It is adaptable for environment of industry production fields or places which is not reachable by people. This technology is featured by strong interference-proof, low-energy-consumption, and real-time communication, which is the functional extension as well as enhancement on industry application of industry communication technology, and also the important complementation and perfection of the field bus control network. WIA-PA project, supported and invested by national 863 plan, was an industrial wireless network technology, which was independently developed by around 10 organizations including Chinese Academy of Sciences and Southwest University. The technical specification of this technology, which got adopted by IEC in 2008, was listed as international standards establishment plan.The underlying network layer of WIA-PA is based on the IEEE 802.15.4. Physical layer is various channels of IEEE 802.15.4, link layer is compatible with superframe structure of IEEE 802.15.4, and has extension according to the requirements of the industrial wireless network. WIA-PA adopts the centralized and distributed integration method to allocate the wireless communication resources. At the mesh layer, gateway intensively allocates resources for routers. At the star layer, router allocates resources for every locale devices. This thesis is focused on the cluster channel resources which gateway intensively allocates for routers at the mesh layer in WIA-PA. Cluster channel is the kind of channel which is used when every cluster runs its internal cluster transactions. In order to avoid interference, keep the reliability of network communicating, let the adjacent clusters use the different cluster channel to communicate. For the same time, we can use the same channel between the nonadjacent clusters which are beyond the distance of communication. Consequently, the problem of the cluster channel allocation of WIA-PA is actually the problem of graph of vertex coloring.During the participation of WIA-PA project, according to the author's deeply analyzing and researching the problem of cluster channel allocation, it can indicate that the existing vertex coloring algorithm cannot satisfy the special requirements of industrial wireless network of WIA-PA. Heuristic and their hybrid algorithms have high complexity and their execution time is long, which can not satisfy the real-time capability of WIA-PA; Welsh-Powell execution efficiency is fast, but it can not adapt to change of network topology structure. So in order to industrial wireless network of WLA-PA, this thesis proposes a simple aggregation theory-based algorithm named SBK-Coloring, of which the concept is that every aggregation maintains nonadjacent vertexes which can have the same color. Each group represents one color. Its cluster channel allocating process is to divide from the graph of vertexes to the independent group. In other words, this is a process of adding vertex to some group. This algorithm complexity is low and its execution efficiency is high. Besides, when the network topology structure changes, it remains original allocating strategy, and avoids a large amount of updating of network information and reconfiguration of network parameters. Additionally, it improves the network capability of real-time and reduces the nodes'energy consumption.The simulation experiments, with analysis and comparison of two typical graph coloring algorithms, result out that execution efficiency of SBK-Coloring algorithm is much higher than the hybrid genetic algorithm. Whether different sparse graphs or different dense graphs, almost all their execution times do not exceed tens of milliseconds. But execution times of hybrid genetic algorithm are or small as several seconds or large as several minutes or tens of minutes. SBK-Coloring algorithm, compared with Welsh-Powell, shows the same capability in the small sparse graphs; But with the increasing of number of vertexes, it shows the obvious advantage. When the vertexes add to 400, it improves by about 47.8%; when the vertexes add to 1000, it improves by about 56.5%. As well, SBK-Coloring shows better capability in the dense graphs. When vertexes add to 95, it improves by about 56.6%; When the number add to 500, it improves by about 88.7%; And when the number add to 1000, it improves by 76.1%. On the whole, in the execution efficiency, SBK-Coloring heads and shoulders above hybrid genetic algorithm, and exceed Welsh-Powell by about 50%in general, In the quality of solutions, SBK-Coloring is a little worse than Welsh-Powell and worse than hybrid genetic algorithm.At last, apply SBK-Coloring to several typical network topology structure of WIA-PA of cluster channel allocation. The result shows that whether in the number of using channel s or execution time of algorithm, it can both meet requirements of WIA-PA.