We investigate several types of Stop and Wait HARQ schemes under imperfect feedback conditions. Performance are measured through the usual metrics, i.e., packet error rate, delay, and efficiency. The main novelties of this paper are twofold: i) the considered feedback undergoes the random arrival time of the acknowledgment. ii) the analysis is done at the network level (IP) which enlarges the vision of practical systems performance. In particular, our analysis enables us to consider a recent cross-layer MAC-IP allocation strategy. Finally our results provide a way to design the feedback link.

In this paper we derive performance metrics closed-form expressions of retransmission schemes such as Automatic Repeat reQuest (ARQ) and Hybrid ARQ (HARQ) in the case of memoryless block fading channels. As performance metrics, we consider the Packet Error Rate (PER), the efficiency, and the packet delay. The novelty of the paper is twofold: i) the metrics are considered at the Network level; ii) we introduce a new general framework which enables us to derive analytically the considered performance metrics for most retransmission schemes, including recent cross-layer strategies with the Network layer. The metrics at the Medium Access Control (MAC) level are obtained as a byproduct. Among the considered retransmission schemes, we especially consider the Incremental Redundancy HARQ scheme with different lengths redundancy packets.

We aim to study a specific cross-layer optimized Hybrid Automatic Retransmission reQuest (HARQ) introduced in [1] and called early-drop HARQ. We remind that in [1], instead of giving a transmission credit per MAC packet (as usually done), it is proposed to share the transmission credit among all the MAC packets belonging to the same IP packet. The early-drop version stops the retransmission as soon as the number of remaining MAC packets is higher than the number of possible transmission attempts. As the early-drop has never been analyzed neither through simulations nor analytical derivations, the purpose of this paper is to fill this gap. As the packet error rate and the delay are not modified, we only focus on the efficiency which is expressed in closed-form.