In this section, as we mainly focus on the local behavior of GOR, for a given transmitter ni, we abbreviate its forwarding candidate set as , and its available next-hop node set as . Note that, is an ordered subset of , which is a set of all the neighbors that are geographically closer to the destination than the transmitter ni. ni's neighbor , its advancement to the destination , and the PRR on link are simplified as iq, dq, and pq, respectively. We assume the packet reception ratios (pq) are independent of each other. The independence has been validated in practice (Laufer and Kleinrock 2008a; Reis et al. 2006). We denote the number of nodes in as r, and the number of nodes in as M. Redefine , and . Note that, the subscript of i only represents the sequence number of each node in set and , and two nodes having the same subscript in and are not necessarily the same node. For example, i1 in does not necessarily indicate the same node as i1 in . Without loss of generality, we assume all the nodes in and are descending ordered according to the advancement s.t. given nodes im and in, we have dm > dn, .
Let be one permutation of nodes in , and the order indicates that nodes will attempt to forward the packet with priority . We define the EPA for the ordered forwarding candidate set in Equation (2.1)
where and . The physical meaning of Equation (2.1) is the expected packet advancement achieved by GOR in one transmission using the ordered forwarding candidate set . The EPA metric accurately indicates the relationship between the packet advancement and candidate selection and prioritization. Note that when r = 1, Equation (2.1) degenerates to the “distance × PRR” proposed in geographic routing (Lee et al. 2005; Seada et al. 2004).