In this section, we will present HAVC protocol, which consists of addressing, encapsulation, connection, and multiple transmission.
8 and 9 illustrate HAVC connection and the corresponding timing diagram, respectively.
HAVC-D is a method of distributing HAVC frames among all available HAVC paths (for example, N paths as shown in Fig.
i] := Maximum HAC frame size/Bandwidth of the i-th HAVC path is the corresponding maximum transmission time, and
Td := Round trip time of the i-th HAVC path/2 is the corresponding end-to-end delay (between the user and HAVC-S), and then HAVC frames are sent through the HAVC path having the lowest score.
The result has shown that latency of HAVC is about 2-4 ms larger than that of IPSec (which is considered as a baseline) and its alternative, namely secure sockets layer (SSL) [32, Chapter 27].
Further, it is worthwhile to evaluate how well the proposed HAVC performs in dynamic networking conditions.
In this subsection, we conduct two HAVC experiments: one on TCP applications and the other on UDP applications.
This measurement indicates that the latency of HAVC is 15-20% more than that of IPSec.
In this subsection, we will investigate the ability of HAVC to enable a user to continuously access the Internet in case of VPN gateway and network failures.
three-path HAVC (Case 1): 1) Connect a user to G/W1, G/W2, and G/W3 for cloud access; 2) Drop one of these VPN gateways randomly at 30 s and recover it at 70 s; and 3) Drop one of the three active gateways randomly at 100 s and recover it at 120 s.
three-path HAVC (Case 2): 1) Connect a user to G/W1, G/W2, and G/W3 for cloud access; 2) Drop two of these VPN gateways randomly at 30 s, and recover one of them at 70 s and the other at 100 s.