This paper addresses the multi-agent consensus problem with an active leader and variable interconnection topology. The leader's state is not necessarily measurable, and each agent only receives the leader's position when connected. To track the leader, a neighbor-based local controller and a neighbor-based state-estimation rule are proposed for each agent. The control scheme ensures that agents can follow the leader if the leader's input is known, and the tracking error can be estimated if the input is unknown. The problem is formulated using graph theory, where the agents and leader are represented as a graph. The graph's topology changes over time, and the agents must adapt to these changes. The leader's dynamics are described by a system with unknown acceleration input, and the agents must estimate the leader's velocity using a dynamic observer. The paper presents a Lyapunov-based approach to analyze the system's stability. A decentralized control scheme is proposed, consisting of a feedback law and a dynamic observer to estimate the leader's velocity. The control law ensures that agents follow the leader, even when the leader's input is unknown. The analysis shows that the tracking error can be bounded, and the system converges to the leader's state if the leader's input is known. The results are extended to cases where the interconnection graph is not always connected. The analysis shows that the system can still converge to the leader's state if the graph remains connected for a sufficient amount of time. The paper also discusses the extension of the results to more complex leader dynamics and provides a framework for distributed observer design in multi-agent systems.This paper addresses the multi-agent consensus problem with an active leader and variable interconnection topology. The leader's state is not necessarily measurable, and each agent only receives the leader's position when connected. To track the leader, a neighbor-based local controller and a neighbor-based state-estimation rule are proposed for each agent. The control scheme ensures that agents can follow the leader if the leader's input is known, and the tracking error can be estimated if the input is unknown. The problem is formulated using graph theory, where the agents and leader are represented as a graph. The graph's topology changes over time, and the agents must adapt to these changes. The leader's dynamics are described by a system with unknown acceleration input, and the agents must estimate the leader's velocity using a dynamic observer. The paper presents a Lyapunov-based approach to analyze the system's stability. A decentralized control scheme is proposed, consisting of a feedback law and a dynamic observer to estimate the leader's velocity. The control law ensures that agents follow the leader, even when the leader's input is unknown. The analysis shows that the tracking error can be bounded, and the system converges to the leader's state if the leader's input is known. The results are extended to cases where the interconnection graph is not always connected. The analysis shows that the system can still converge to the leader's state if the graph remains connected for a sufficient amount of time. The paper also discusses the extension of the results to more complex leader dynamics and provides a framework for distributed observer design in multi-agent systems.