Optimizing Network Measurements through Self-adaptive Sampling

Traffic sampling techniques are crucial and extensively used to assist network management tasks. Nevertheless, combining accurate network parameters' estimation and flexible lightweight measurements is an open challenge. In this context, this paper proposes a self-adaptive sampling technique, based on linear prediction, which allows to reduce significantly the measurement overhead, while assuring that sampled traffic reflects the statistical characteristics of the global traffic under analysis. The technique is multiadaptive as several parameters are considered in the dynamic configuration of the traffic selection process. The devised test scenarios aim at exploring the proposed sampling technique ability to join accurate network estimates to reduced overhead, using throughput as reference parameter. The evaluation results, obtained resorting to real traffic traces representing wired and wireless aggregated traffic scenarios and actual network services, prove that the simplicity, flexibility and self-adaptability of this technique can be successfully explored to improve network measurements efficiency over distinct traffic conditions. For optimization purposes, this paper also includes a study of the impact of varying the order of prediction, i.e., of considering different degrees of past memory in the self-adaptive estimation mechanism. The significance of the obtained results is demonstrated through statistical benchmarking.