In today’s computer networks, short-lived flows are predominant. Consequently,transient start-up effects such as the connection establishment incellular networks have a significant impact on the performance. Althoughvarious solutions are derived in the fields of queuing theory, available bandwidths,and network calculus, the focus is, e.g., about the mean wake-uptimes, estimates of the available bandwidth, which consist either out of asingle value or a stationary function and steady-state solutions for backlogand delay. Contrary, the analysis during transient phases presents fundamentalchallenges that have only been partially solved and is thereforeunderstood to a much lesser extent.To better comprehend systems with transient characteristics and to explaintheir behavior, this thesis contributes a concept of non-stationaryservice curves that belong to the framework of stochastic network calculus.Thereby, we derive models of sleep scheduling including time-variantperformance bounds for backlog and delay. We investigate the impact ofarrival rates and different duration of wake-up times, where the metricsof interest are the transient overshoot and relaxation time. We comparea time-variant and a time-invariant description of the service with anexact solution. To avoid probabilistic and maybe unpredictable effects fromrandom services, we first choose a deterministic description of the serviceand present results that illustrate that only the time-variant service curve canfollow the progression of the exact solution. In contrast, the time-invariantservice curve remains in the worst-case value.Since in real cellular networks, it is well known that the service and sleepscheduling procedure is random, we extend the theory to the stochasticcase and derive a model with a non-stationary service curve based onregenerative processes.Further, the estimation of cellular network’s capacity/ available bandwidthfrom measurements is an important topic that attracts research, andseveral works exist that obtain an estimate from measurements. Assuminga system without any knowledge about its internals, we investigateexisting measurement methods such as the prevalent rate scanning andthe burst response method. We find fundamental limitations to estimatethe service accurately in a time-variant way, which can be explained bythe non-convexity of transient services and their super-additive networkprocesses.In order to overcome these limitations, we derive a novel two-phase probingtechnique. In the first step, the shape of a minimal probe is identified,which we then use to obtain an accurate estimate of the unknown service.To demonstrate the minimal probing method’s applicability, we performa comprehensive measurement campaign in cellular networks with sleepscheduling (2G, 3G, and 4G). Here, we observe significant transient backlogsand delay overshoots that persist for long relaxation times by sendingconstant-bit-rate traffic, which matches the findings from our theoreticalmodel. Contrary, the minimal probing method shows another strength:sending the minimal probe eliminates the transient overshoots and relaxationtimes.