Initially the robot has pain w, set according to the quality of the work to be performed, and it has an empty RRT buffer. It keeps on performing its task with the aim to minimize the pain w. If the task requires any information to be sent the pain c is incremented. Then the robot starts sensing for the communication signal from its neighbors. If no signal is received the robot starts scanning for neighbors. If no neighbor units are around the robot will take one RRT step, through RRT forward propagation block in Fig. 2, and push the initial location node information into the RRT buffer as shown in Table 1. By doing so the robot has acquired an increment in the pain c. More c represents that ...view middle of the document...
Moreover, for dynamic networks where node topology is not fixed and new nodes are added frequently as well as existing nodes can lose connection with the network, updating of routing table may result in high consumption of computation and memory. To overcome such a problem the use of Gossip algorithm is preferred. Gossip algorithms  are randomized methods designed to transmit a message from a source to a destination without any explicit route discovery mechanisms.
If two nodes are within communication range, we say that they are neighbors. When a robot wants to send a message to a destination robot, it randomly chooses one of its neighbors and transmits the message together with a header that contains the identifier of the destination and itself, and a time value indicating the validity period of the message. In an alternative approach of the protocol, the message is simply broadcast to all neighbors, but depending on the application, this may incur high communication load in the network.
For a population size of n, assuming that the individuals try to send packets to f other members in each round and no packet is discarded, the fractional penetration of the packet Yr in round r  is given by -
Here maximum value of r is the time to live (TTL) of the packet. We can calculate the optimum value of TTL using this. For various value of ‘f’, the variation of fractional penetration Yr with number of rounds r is given by Fig. 3.
We in our system use a modification in this algorithm to minimize the load on the robots and reduce the delay incurred during routing. In this system the robot broadcasts its information to all the neighbors and then waits for the acknowledgment, this can be seen in the Transmission block in Fig. 2. If any robot senses this communication signal, first compare its own address to the destination address from header of the message. If a match is found the robot broadcasts the acknowledgment among its neighbors and modifies its w value according to the importance of message in the task performed by the destination robot. If the match is not found the receiver robot compares the w, c value of the sender robot. It can be noted that the w, c value is also required to be present in the header information. If the value is higher than its own then also the receiver broadcasts the acknowledgment and increases its c value. This is done in the Signal Interpretation Block as shown in Fig. 2.
In our modified version of gossip algorithm we forward a packet only when the pain of the receiving robot in terms of w and c is less than the pain of sending robot. Hence f depends on the equivalent pains of both the participating robots, Xr and Xs, and is given by (2)
Here α, β are the predefined weights of w and c which depends upon the task being performed by the robot. Since the value of f varies on each new location node that the robot traverses therefore to calculate the Fractional penetration...