Solving TSP based on Multi-Segment Multi-Orientation Nearest Neighbor algorithm

Considering constructional algorithms for solving Traveling Salesman Problem (TSP), the solution process of Nearest Neighbor (NN) is the simplest one, but its performance is the worst. This paper proposes an improved NN algorithm for solving TSP in Euclidean Plane. Firstly, the algorithm searches four corner-nodes from all nodes, and gets a simple tour which only includes four corner-nodes, obtains four segments and their start nodes and end nodes, respectively. Then, each segment starts from current corner-node, and the algorithm selects a node as next target node from its multi-orientation nearest neighbor nodes, considering their distance and orientation comprehensively, and searching process repeatedly until each segment arrives at their end nodes. Furthermore, the algorithm inserts those rest nodes into a certain segment. At last, the algorithm connects four segments from stem to stern to form the final tour. Experimental results indicate that the proposed algorithm greatly improves the result of NN algorithm.